HUAWEI Enterprise Solution White Paper

Transcription

1 HUAWEI Enterprise Solution White Paper Issue 01 Date HUAWEI TECHNOLOGIES CO., LTD.

2 All rights reserved. No part of this document may be reproduced or transmitted in any form or by any means without prior written consent of Huawei Technologies Co., Ltd. Trademarks and Permissions and other Huawei trademarks are trademarks of Huawei Technologies Co., Ltd. All other trademarks and trade names mentioned in this document are the property of their respective holders. Notice The purchased products, services and features are stipulated by the commercial contract made between Huawei and the customer. All or partial products, services and features described in this document may not be within the purchased scope or the usage scope. Unless otherwise agreed by the contract, all statements, information, and recommendations in this document are provided "AS IS" without warranties, guarantees or representations of any kind, either express or implied. The information in this document is subject to change without notice. Every effort has been made in the preparation of this document to ensure accuracy of the contents, but all statements, information, and recommendations in this document do not constitute a warranty of any kind, express or implied. Huawei Technologies Co., Ltd. Address: Website: Huawei Industrial Base Bantian, Longgang Shenzhen People's Republic of China support@huawei.com Telephone: Fax: i

3 Contents Contents 1 Overview Background Overall Requirements for a Mail System Overview of HUAWEI Enterprise Solution Highlights of HUAWEI Enterprise Solution Architecture of HUAWEI Enterprise Solution Introduction to the Exchange Server Design of HUAWEI Enterprise Solution System Design Rules Low-level Design Network Topologies of Huawei Exchange Solutions MBS Structure Server Performance Design Server Performance Design for the Low-cost 600-user Solution Server Performance Design for the Low-cost 1200-user Solution Server Performance Design for the High-performance 1200-user Solution Server Performance Design for the Low-cost 3000-user Solution MBS Storage Design MBS Storage Design for the Low-cost 600-user Solution MBS Storage Design for the Low-cost 1200-user Solution MBS Storage Design for the High-performance 1200-user Solution MBS Storage Design for the Low-cost 3000-user Solution AD Design Antispam and Antivirus Design Design Rules Antispam and Antivirus Requirements Networking Scheme and Design Load Balancing Scheme Design Hardware Load Balancing Scheme Backup Scheme Design Overall Requirements for the Backup System ii

4 Contents Networking for the Backup Scheme Backup Capacity Calculation Performance Design for the Backup Scheme Archiving Scheme Design Overall Requirements for the Archiving System Networking for the Archiving Scheme Archiving Capacity Calculation Design for Forefront TMG Overall Requirements Overall Scheme Design Architecture Design and Network Topology HA and Scalability Design Publishing Design Web Access Design Array Design LB Design Protection Design Network Design DAG Network Configuration of the Exchange Server HA and Disaster Recovery Restrictions on Mailboxes Management Mode Client Access Mode System Security Configuration Lists Configuration List for the Low-cost 600-user Solution Hardware Configurations Software Configurations Link Configurations Configuration List for the Low-cost 1200-user Solution Hardware Configurations Software Configurations Link Configurations Configuration List for the High-performance 1200-user Solution Hardware Configurations Software Configurations Link Configurations Configuration List for the Low-cost 3000-user Solution Hardware Configurations Software Configurations iii

5 Contents Link Configurations Involved Products Huawei RH2285 V Quantum Tape Library F5 BIG-IP LTM Acronyms and Abbreviations iv

6 1 Overview 1 Overview 1.1 Background Nowadays, the mail system has become a core component of the information system for a company. It is the pivot for office automation and service transfer within a company or between the company and external systems. Company A has about N employees and the employee base will increase to about 3000 in the coming three years. Company A has no mail system or Active Directory (AD) service deployed. As the company business develops, more function and security requirements are posed for the information system. This urges company A to establish a mail system. 1.2 Overall Requirements for a Mail System A mail system provides internal mail services, external mail services, connection services for external users, and public folders. When the Internet is unavailable, it ensures that mails can be properly exchanged within a company. Functional Requirements for a Mail System The mail system is expected to provide the following functions: Access service: The mail system supports multiple access modes such as Outlook, Foxmail, and mainstream Web browsers over protocols like the Simple Mail Transfer Protocol (SMTP), Post Office Protocol 3 (POP3), and the Interactive Mail Access Protocol 4 (IMAP4). Antispam: The mail system filters mails by connection, recipient, sender, sender ID, content, attachment, or sender credit. It converges mails on the server based on specified rules. Antivirus: Third-party multiengine antivirus software is embedded into or referenced by the mail system. The antivirus software filters mail attachments based on their headers. The antivirus software detects and clears viruses for both mails and SMTP servers. Impeccable management and monitoring capabilities: The mail system supports multilevel authorization. It traces and monitors mails. User classification: The mail system allows you to set user-specific client access modes, mail storage restrictions, and mail transfer restrictions. 1

7 1 Overview Basic Functions Customization support: Interfaces are provided and can be invoked by other applications for customization. Personal information management: The mail system provides common personal information management functions including contacts management, schedule management, and task management. Capacity alerting: The mail system notifies users of the usage of their configured mailbox space. The mail system provides the following functions for individual users: Allows users to access mailboxes by using the Internet Explorer. Receives, sends, replies to, transfers, deletes, and signs mails, and sets automatic reply. Manages the received mails, sent mails, drafts, and spam mails. Defines personal folders. Defines personal address books. Filters mails. Sets automatic archiving. Accesses the mailbox remotely. Defines a public address book. Supports compound texts so that mails can contain tables, figures, images, audio clips, animation, and multimedia information. The mail system also supports security mechanisms including digital signature and encryption. Allows a user to manage and access any information such as a mail, schedule, contact, or task from any client. 1.3 Overview of HUAWEI Enterprise Solution Company A requires a mail system that provides basic mail functions, high availability (HA), high reliability, as well as powerful communication functions. Such a mail system can intelligently associate various communication modes for company A. Microsoft Exchange Server 2010, which provides high reliability, a user-friendly user interface (UI), and multiple access modes, is the best choice. Microsoft Exchange Server is a versatile application server for collaboration on an intranet. It provides industry-leading scalability, reliability, security, and performance. It provides a wide range of collaborative applications including , meeting arrangement, schedule management for teams, task management, document management, real-time conferencing, and workflows. All applications can be accessed from the Internet by using a browser. Huawei customizes an Exchange solution for company A based on the Exchange Server The HUAWEI Enterprise Solution is a one-stop solution that encompasses both hardware and software. The solution involves creating an AD that bases on Windows Server 2008 R2 and deploying the Exchange Server The Exchange Server 2010 mail system takes a leading position in the industry for high performance, reliability, security, and scalability. The solution involves: Planning and deploying the hardware platform for HUAWEI Enterprise Solution 2

8 1 Overview Deploying the AD service on a Windows Server 2008 R2 server Deploying the Exchange Server 2010 Deploying antispam and antivirus schemes Deploying a system protection scheme Deploying access agents Deploying a load balancing scheme Deploying a mail backup scheme Deploying a mail archiving scheme 1.4 Highlights of HUAWEI Enterprise Solution The highlights of HUAWEI Enterprise Solution are as follows: Ease of use: It can be fast deployed as required to provide a high-availability information platform, improving working efficiency. When more mailboxes are required, increase the storage capacity or add servers using the block design method and add the servers to the cluster. No complex storage configuration is involved. Low cost: Huawei RH2285 V2 configured with Serial Advanced Technology Attachment (SATA) disks is used. This reduces the deployment cost of the user data center and meets capacity and performance requirements. High performance: The solution provides high input/output (I/O) performance and short delay based on solid state disk (SSD) cards developed by Huawei, meeting requirements of high-end users. Integrated mail solution: The solution integrates the RH2285 V2 with the Exchange Server It is a composite service solution that encompasses both software and hardware, thereby lowering the management and maintenance costs. 3

9 2.1 Architecture of HUAWEI Enterprise Solution Figure 2-1 shows the system architecture of HUAWEI Enterprise Solution. Figure 2-1 Architecture of HUAWEI Enterprise Solution ETS (antispam and antivirus) HTS External mails Mobile user MBS Backup server Browser user TMG server (access agent) CAS Outlook (remote user) Archiving server Tape library As shown in Figure 2-1, the HUAWEI Enterprise Solution system comprises the following modules: AD & DNS module Exchange Server 2010 Antispam and antivirus module System protection module Access agent module Mail backup module Mail archiving module 1

10 2.2 Introduction to the Exchange Server 2010 The Exchange Server 2010 is a role-based program. It provides a platform with abundant powerful functions for Internet and intranet mail solutions. Figure 2-2 shows the architecture of the Exchange Server Figure 2-2 Architecture of the Exchange Server 2010 The Exchange Server 2010 involves four server roles: Edge Transport Server (ETS) Hub Transport Server (HTS) Client Access Server (CAS) Mailbox Server (MBS) The HTS, CAS, and MBS can be installed on the same server or independent servers. For security purposes, ETSs cannot be added to domains but installed on the same server or independent servers. This design yields flexible server networking and therefore improves scalability, security and usability. All mails are transferred through the HTS. Filter rules, automatic backup, exemption declaration can be deployed conveniently. All traffic is encrypted within an organization using the Transport Layer Security (TLS) to ensure mail security. The ETS needs to be designed independently. It runs independent of domains. You are advised to deploy the ETS in the demilitarized zone (DMZ) to further harden system security. 2

11 2.3 Design of HUAWEI Enterprise Solution System Design Rules The HUAWEI Enterprise Solution system is designed in line with the following rules: Stability System stability must be ensured from the architecture aspect. The impact of system configurations on stability must also be considered to avoid single point of failure (SPOF). The system must pass a series of strict tests, including functional tests, tests in different environments, and tests under different configurations. This is to ensure that the system can run stably under different circumstances. Ease of use The system must be easy to operate and provide comprehensive and particular prompts, error reports, and monitoring information during operation. Easy management The system platform must be easy to manage to reduce the management workload. Training courses, help documents, and even operation wizards need to be provided to facilitate system management. Easy management must be considered in design of an information exchange system. A user-friendly UI must be provided to guide users through data exchange management. Easy maintenance The system must also be easy to maintain. The functional components of the system run independent of each other to some extent to ensure that a component can be replaced without affecting other components and can work with other components to provide required functions. Scalability The system must be easy to extend in service functions and information exchange regulations to fit in with service development. Security Low-level Design Security settings of the system platform must be used to ensure overall system security. Ensure that the system is protected against unauthorized access, data is not lost during transmission, the system can correctly identify senders and recipients, and data is not obtained illegitimately or tampered. The low-level design is as follows: According to Microsoft's experience, an enterprise mailbox receives and sends 200 mails on average per day, with an average mail size of 75 KB. It is recommended that deleted mails be retained for 14 days. The mailbox databases of the Exchange Server 2010 provide HA and each has two copies including the active database copy. According to Microsoft's experience, the read/write ratio is 3:2 for enterprises that have the Exchange Server 2010 deployed. Configuration for the Low-cost 600-user Exchange Solution Table 2-1 lists the configuration for the low-cost 600-user Exchange solution. 3

12 Table 2-1 Configuration for the low-cost 600-user Exchange solution Configuration Item Value Mailbox quantity 600 Mails received and sent per mailbox per day 200 Average mail size (KB) 75 Mailbox size limit (MB) 1536 Retention period of deleted mails (days) 14 Number of database availability groups (DAGs) 1 Number of MBSs per DAG 2 Number of active databases per server 1 Number of copies per database 2 including the active database copy Number of mailboxes per database 300 Read/write ratio 3:2 Configuration for the Low-cost 1200-User Exchange Solution Table 2-2 lists the configuration for the low-cost 1200-user Exchange solution. Table 2-2 Configuration for the low-cost 1200-user Exchange solution Configuration Item Value Mailbox quantity 1200 Mails received and sent per mailbox per day 200 Average mail size (KB) 75 Mailbox size limit (MB) 3072 Retention period of deleted mails (days) 14 Number of DAGs 1 Number of MBSs per DAG 2 Number of active databases per server 2 Number of copies per database 2 including the active database copy Number of mailboxes per database 300 Read/write ratio 3:2 4

13 Configuration for the High-performance 1200-User Exchange Solution Table 2-3 lists the configuration for the high-performance 1200-user Exchange solution. Table 2-3 Configuration for the high-performance 1200-user Exchange solution Configuration Item Value Mailbox quantity 1200 Mails received and sent per mailbox per day 500 Average mail size (KB) 75 Mailbox size limit (MB) 307 Retention period of deleted mails (days) 7 Number of DAGs 1 Number of MBSs per DAG 2 Number of active databases per server 1 Number of copies per database 2 including the active database copy Number of mailboxes per database 600 Read/write ratio 3:2 Configuration for the Low-cost 3000-User Exchange Solution Table 2-4 lists the configuration for the low-cost 3000-user Exchange solution. Table 2-4 Configuration for the low-cost 3000-user Exchange solution Configuration Item Value Mailbox quantity 3000 Mails received and sent per mailbox per day 200 Average mail size (KB) 75 Mailbox size limit (MB) 3072 Retention period of deleted mails (days) 14 Number of DAGs 1 Number of MBSs per DAG 3 Number of active databases per server 3 or 4 Number of copies per database 3 including the active database copy Number of mailboxes per database 300 5

14 Configuration Item Value Read/write ratio 3:2 According to Table 2-4, a mailbox receives and sends 200 mails on average per day, with the average mail size 75 KB. According to Table 2-5, the average input/output operations per second (IOPS) of a single mailbox is 0.2, as shown in the row highlighted in bold. Table 2-5 Cache and IOPS of each mailbox Number of Mails Received/Sent per Mailbox per Day (Average Mail Size75 KB) Database Cache per Mailbox (MB) Single Database Copy: Estimated IOPS per Mailbox Multiple Database Copies: Estimated IOPS per Mailbox Network Topologies of Huawei Exchange Solutions Network Topology of the 600-user Solution Figure 2-3 shows the network topology of the 600-user solution. 6

15 Figure 2-3 Network topology of the 600-user solution The 600-user Exchange system consists of the CAS, HTS, MBS, ETS, AD & DNS controller, backup server, archiving server, tape library, and system protection software. The MBS, HTS, and CAS are core servers. The 600-user Exchange system uses the redundancy architecture. When one server is faulty, its services are automatically switched over to other normal servers within 30 seconds without affecting customer services. The system is deployed in the data center of company A. The data center is divided into two zones: DMZ and APP & database zone. After a firewall is installed, external systems cannot access servers on the internal network. The DMZ is set as a buffer between insecure systems and a secure system to resolve this problem. The buffer resides in a small network area between the internal network of company A and external networks. Public servers can be deployed in this area. In this solution, the ETSs, Forefront Threat Management Gateway (TMG) servers, and manager server are deployed in the DMZ and other servers are deployed in the APP & Database zone. The components of the 600-user Exchange system are described as follows: Two RH2285 V2 servers for deploying the MBS, HTS, and CAS The Exchange Server 2010 DAG technology is used to construct a mail platform with cluster architecture. Each mailbox database in the DAG has an active copy and a passive copy to implement full redundancy of Exchange services and data structures. When a server (database) is faulty, services are switched over to the other server. Each MBS supports a maximum of 1200 users. This ensures that mail services are not affected when an MBS becomes faulty. The HTS routes all mails across the internal network. When the HTS is faulty, mail receiving and transfer are unavailable. To avoid this problem, two HTSs in redundancy mode are deployed. When an HTS is faulty, the MBS connects to the other HTS. The HTS also controls mail flows within the company to implement virus detection, content-based filtering, and mail archiving, thereby reducing information leakage risks. 7

16 The CAS allows mail clients as well as Web and mobile users to access the MBS. The CAS employs the advanced F5 load balancing technology to implement load balancing and redundancy. The MBS, HTS, and CAS are deployed on the same server to reduce resource usage. Two ETSs Two ETSs are deployed to isolate the internal mail system from external mail systems. When a server connected to external mail systems is attacked by hackers, servers in the APP & Database zone behind the firewall are not affected. This improves security and design flexibility. The ETS also filters out virus-infected mails and spam mails from the Internet. In the domain name service (DNS) round-robin mechanism, the two ETSs provide services simultaneously in load balancing mode. The DNS round-robin mechanism is used to implement network resource sharing and allocation. Two AD & DNS controllers The Windows Server 2008 R2 server or existing AD & DNS controller is used. The DNS service runs on both servers. The running of the AD and DNS services determine the status of the Exchange Server At least two servers must be configured for deploying two AD & DNS controllers in active/standby mode. An AD can be deployed on a 32-bit or 64-bit server, but the functional level of the AD & DNS controllers must be Windows Server 2003 or later to gain higher performance. Four F5 load balancers Two F5 BIG-IP LTM 1600 load balancers are used by CAS. An F5 BIG-IP LTM 1600 can meet the performance requirements in a 1200-user Exchange system. For reliability purposes, two LTM 1600 load balancers in redundancy mode are configured. All clients access the CAS through the LTM Multiple CAS clusters are configured to connect clients. Other two F5 BIG-IP LTM 1600 load balancers are used by TMG server. One backup server In the backup scheme, Huawei's server is the hardware and Symantec NetBackup (NBU) is the software. The NBU is extended for the Exchange Server 2010 so that it can back up the mailbox databases of the Exchange Server Two archiving servers and two tape libraries The mail archiving module comprises the archiving server, database server (SQL Server), tape library, and Symantec's archiving software Enterprise Vault. Mails are automatically archived and managed. Both mails and mail logs are archived. Archiving policies can be easily configured to reduce space occupation on the active disk, thereby ensuring that archived mails meet verification requirements. Two TMGs The TMG module authenticates Internet users that access Huawei Exchange system by using Outlook Web App or Outlook Anywhere, and only authenticated users are allowed based on specified policies. Internal users can access the Internet through the TMG module. In the solution, two RH2285 V2 servers are configured in load balancing mode and the software is Microsoft's Forefront TMG. One manager server The manager server is responsible for system protection. System protection involves protecting the operating system against viruses and ensuring security of servers. In the solution, the mainstream Symantec Endpoint Protection (SEP) antivirus software is used. An SEP client is deployed for each server. The Symantec Endpoint Protection Manager (SEPM) is installed on the manager server. The manager server manages all SEP clients on the local area network (LAN). That is, software deployment, policy configuration, and engine upgrade are completed on the manager server. 8

17 Network Topology of the Low-cost 1200-user Solution Figure 2-4 shows the network topology of the low-cost 1200-user solution. Figure 2-4 Network topology of the low-cost 1200-user solution The 1200-user Exchange system contains the CAS, HTS, MBS, ETS, AD & DNS controller, backup server, archiving server, tape library, hardware load balancer, and system protection software. The MBS, HTS, and CAS are core servers. In the solution, they are installed on the same server. The 1200-user Exchange system uses the redundancy architecture. When one server is faulty, its services are automatically switched over to other normal servers within 30 seconds without affecting customer services. The system is deployed in the data center of company A. The data center is divided into two zones: DMZ and APP & database zone. After a firewall is installed, external systems cannot access servers on the internal network. The DMZ is set as a buffer between insecure systems and a secure system to resolve this problem. The buffer resides in a small network area between the internal network of company A and external networks. Public servers can be deployed in this area. In this solution, the ETSs, TMG servers, and manager server are deployed in the DMZ and other servers are deployed in the APP & Database zone. The components of the low-cost 1200-user Exchange system are described as follows: Two RH2285 V2 servers for deploying the MBS, HTS, and CAS The Exchange Server 2010 DAG technology is used to construct a mail platform with cluster architecture. Each mailbox database in the DAG has an active copy and a passive copy to implement full redundancy of Exchange services and data structures. When a server (database) is faulty, services are switched over to the other server. Each MBS supports a maximum of 1200 users. This ensures that mail services are not affected when an MBS becomes faulty. The HTS routes all mails across the internal network. When the HTS is faulty, mail receiving and transfer are unavailable. To avoid this problem, two HTSs in redundancy mode are deployed. When an HTS is faulty, the MBS connects to the other HTS. The 9

18 HTS also controls mail flows within the company to implement virus detection, content-based filtering, and mail archiving, thereby reducing information leakage risks. The CAS allows mail clients as well as Web and mobile users to access the MBS. The CAS employs the advanced F5 load balancing technology to implement load balancing and redundancy. The MBS, HTS, and CAS are deployed on the same server to reduce resource usage. Two ETSs Two ETSs are deployed to isolate the internal mail system from external mail systems. When a server connected to external mail systems is attacked by hackers, servers in the APP & Database zone behind the firewall are not affected. This improves security and design flexibility. The ETS also filters out virus-infected mails and spam mails from the Internet. In the DNS round-robin mechanism, the two ETSs provide services simultaneously in load balancing mode. The DNS round-robin mechanism is used to implement network resource sharing and allocation. Two AD & DNS controllers The Windows Server 2008 R2 server or existing AD & DNS controller is used. The DNS service runs on both servers. The running of the AD and DNS services determine the status of the Exchange Server At least two servers must be configured for deploying two AD & DNS controllers in active/standby mode. An AD can be deployed on a 32-bit or 64-bit server, but the functional level of the AD & DNS controllers must be Windows Server 2003 or later to gain higher performance. Four F5 load balancers Two F5 BIG-IP LTM 1600 load balancers are used by CAS. An F5 BIG-IP LTM 1600 can meet the performance requirements in a 1200-user Exchange system. For reliability purposes, two LTM 1600 load balancers in redundancy mode are configured. All clients access the CAS through the LTM Multiple CAS clusters are configured to connect clients. Other two F5 BIG-IP LTM 1600 load balancers are used by TMG server. One backup server In the backup scheme, Huawei's server is the hardware and Symantec NBU is the software. The NBU is extended for the Exchange Server 2010 so that it can back up the mailbox databases of the Exchange Server Two archiving servers and two tape libraries The mail archiving module comprises the archiving server, database server (SQL Server), tape library, and Symantec's archiving software Enterprise Vault. Mails are automatically archived and managed. Both mails and mail logs are archived. Archiving policies can be easily configured to reduce space occupation on the active disk, thereby ensuring that archived mails meet verification requirements. Two TMGs The TMG module authenticates Internet users that access Huawei Exchange system by using Outlook Web App or Outlook Anywhere, and only authenticated users are allowed based on specified policies. Internal users can access the Internet through the TMG module. In the solution, two RH2285 V2 servers are configured in load balancing mode and the software is Microsoft's Forefront TMG. One manager server The manager server is responsible for system protection. System protection involves protecting the operating system against viruses and ensuring security of servers. In the solution, the mainstream SEP antivirus software is used. An SEP client is deployed for each server. The SEPM is installed on the manager server. The manager server manages 10

19 all SEP clients on the LAN. That is, software deployment, policy configuration, and engine upgrade are completed on the manager server. Network Topology of the High-performance 1200-user Solution Figure 2-5 shows the network topology of the high-performance 1200-user solution. Figure 2-5 Network topology of the high-performance 1200-user solution The 1200-user Exchange system contains the CAS, HTS, MBS, ETS, AD & DNS controller, backup server, archiving server, tape library, hardware load balancer, and system protection software. The MBS, HTS, and CAS are core servers. In the solution, these three roles are deployed on the same server. The 1200-user Exchange system uses the redundancy architecture. When one server is faulty, its services are automatically switched over to other normal servers within 30 seconds without affecting customer services. The system is deployed in the data center of company A. The data center is divided into two zones: DMZ and APP & database zone. After a firewall is installed, external systems cannot access servers on the internal network. The DMZ is set as a buffer between insecure systems and a secure system to resolve this problem. The buffer resides in a small network area between the internal network of company A and external networks. Public servers can be deployed in this area. In this solution, the ETSs, TMG servers, and manager server are deployed in the DMZ and other servers are deployed in the APP & Database zone. The components of the high-performance 1200-user Exchange system are described as follows: Two RH2285 V2 servers for deploying the MBS, HTS, and CAS The Exchange Server 2010 DAG technology is used to construct a mail platform with cluster architecture. Each mailbox database in the DAG has an active copy and a passive copy to implement full redundancy of Exchange services and data structures. When a server (database) is faulty, services are switched over to the other server. Each MBS supports a maximum of 1200 users. This ensures that mail services are not affected when an MBS becomes faulty. 11

20 The HTS routes all mails across the internal network. When the HTS is faulty, mail receiving and transfer are unavailable. To avoid this problem, two HTSs in redundancy mode are deployed. When an HTS is faulty, the MBS connects to the other HTS. The HTS also controls mail flows within the company to implement virus detection, content-based filtering, and mail archiving, thereby reducing information leakage risks. The CAS allows mail clients as well as Web and mobile users to access the MBS. The CAS employs the advanced F5 load balancing technology to implement load balancing and redundancy. The MBS, HTS, and CAS are deployed on the same server to reduce resource usage. Two ETSs Two ETSs are deployed to isolate the internal mail system from external mail systems. When a server connected to external mail systems is attacked by hackers, servers in the APP & Database zone behind the firewall are not affected. This improves security and design flexibility. The ETS also filters out virus-infected mails and spam mails from the Internet. In the DNS round-robin mechanism, the two ETSs provide services simultaneously in load balancing mode. The DNS round-robin mechanism is used to implement network resource sharing and allocation. Two AD & DNS controllers The Windows Server 2008 R2 server or existing AD & DNS controller is used. The DNS service runs on both servers. The running of the AD and DNS services determine the status of the Exchange Server At least two servers must be configured for deploying two AD & DNS controllers in active/standby mode. An AD can be deployed on a 32-bit or 64-bit server, but the functional level of the AD & DNS controllers must be Windows Server 2003 or later to gain higher performance. Four F5 load balancers Two F5 BIG-IP LTM 1600 load balancers are used by CAS. An F5 BIG-IP LTM 1600 can meet the performance requirements in a 1200-user Exchange system. For reliability purposes, two LTM 1600 load balancers in redundancy mode are configured. All clients access the CAS through the LTM Multiple CAS clusters are configured to connect clients. Other two F5 BIG-IP LTM 1600 load balancers are used by TMG server. One backup server In the backup scheme, Huawei's server is the hardware and Symantec NBU is the software. The NBU is extended for the Exchange Server 2010 so that it can back up the mailbox databases of the Exchange Server Two archiving servers and two tape libraries The mail archiving module comprises the archiving server, database server (SQL Server), tape library, and Symantec's archiving software Enterprise Vault. Mails are automatically archived and managed. Both mails and mail logs are archived. Archiving policies can be easily configured to reduce space occupation on the active disk, thereby ensuring that archived mails meet verification requirements. Two TMGs The TMG module authenticates Internet users that access Huawei Exchange system by using Outlook Web App or Outlook Anywhere, and only authenticated users are allowed based on specified policies. Internal users can access the Internet through the TMG module. In the solution, two RH2285 V2 servers are configured in load balancing mode and the software is Microsoft's Forefront TMG. One manager server The manager server is responsible for system protection. System protection involves protecting the operating system against viruses and ensuring security of servers. In the 12

21 solution, the mainstream SEP antivirus software is used. An SEP client is deployed for each server. The SEPM is installed on the manager server. The manager server manages all SEP clients on the LAN. That is, software deployment, policy configuration, and engine upgrade are completed on the manager server. Network Topology of the Low-cost 3000-user Solution Figure 2-6 shows the network topology of the low-cost 3000-user solution. Figure 2-6 Network topology of the low-cost 3000-user solution The 3000-user Exchange system contains the CAS, HTS, MBS, ETS, AD & DNS controller, backup server, archiving server, tape library, hardware load balancer, and system protection software. The MBS, HTS, and CAS are core servers. In the solution, these three roles are deployed on the same server. The 3000-user Exchange system uses the redundancy architecture. When one server is faulty, its services are automatically switched over to other normal servers within 30 seconds without affecting customer services. The system is deployed in the data center of company A. The data center is divided into two zones: DMZ and APP & database zone. After a firewall is installed, external systems cannot access servers on the internal network. The DMZ is set as a buffer between insecure systems and a secure system to resolve this problem. Public servers can be deployed in this area. In this solution, the ETSs, TMG servers, and manager server are deployed in the DMZ and other servers are deployed in the APP & Database zone. The components of the low-cost 3000-user Exchange system are described as follows: Three Huawei RH2285 V2 servers for deploying the MBS, HTS, and CAS The Exchange Server 2010 DAG technology is used to construct a mail platform with cluster architecture. The DAG contains one active and two standby mail databases to implement full redundancy for mail exchange services and data structures. When a server (database) is faulty, mail services are switched over to a standby server. Each MBS 13

22 supports a maximum of 3000 users. This ensures that mail services are not affected when an MBS becomes faulty. The HTS routes all mails across the internal network. When the HTS is faulty, mail receiving and transfer are unavailable. To avoid this problem, three servers in redundancy mode are deployed. When an HTS is faulty, the MBS connects to the other HTS. The HTS also controls mail flows within the company to implement virus detection, content-based filtering, and mail archiving, thereby reducing information leakage risks. The CAS allows mail clients as well as Web and mobile users to access the MBS. The CAS employs the advanced F5 load balancing technology to implement load balancing and redundancy. The MBS, HTS, and CAS are deployed on the same server to reduce resource usage. Two ETSs Two ETSs are deployed to isolate the internal mail system from external mail systems. When a server connected to external mail systems is attacked by hackers, servers in the APP & Database zone behind the firewall are not affected. This improves security and design flexibility. The ETS also filters out virus-infected mails and spam mails from the Internet. In the DNS round-robin mechanism, the two ETSs provide services simultaneously in load balancing mode. The DNS round-robin mechanism is used to implement network resource sharing and allocation. Two AD & DNS controllers The Windows Server 2008 R2 server or existing AD & DNS controller is used. The DNS service runs on both servers. The running of the AD and DNS services determine the status of the Exchange Server At least two servers must be configured for deploying two AD & DNS controllers in active/standby mode. An AD can be deployed on a 32-bit or 64-bit server, but the functional level of the AD & DNS controllers must be Windows Server 2003 or later to gain higher performance. Four F5 load balancers Two F5 BIG-IP LTM 3600 load balancers are used by CAS. An F5 BIG-IP LTM 1600 can meet the performance requirements in a 1200-user Exchange system. For reliability purposes, two LTM 1600 load balancers in redundancy mode are configured. All clients access the CAS through the LTM Multiple CAS clusters are configured to connect clients. Other two F5 BIG-IP LTM 1600 load balancers are used by TMG server. One backup server In the backup scheme, Huawei's server is the hardware and Symantec NBU is the software. The NBU is extended for the Exchange Server 2010 so that it can back up the mailbox databases of the Exchange Server Two archiving servers and two tape libraries The mail archiving module comprises the archiving server, database server (SQL Server), tape library, and Symantec's archiving software Enterprise Vault. Mails are automatically archived and managed. Both mails and mail logs are archived. Archiving policies can be easily configured to reduce space occupation on the active disk, thereby ensuring that archived mails meet verification requirements. Two TMGs The TMG module authenticates Internet users that access Huawei Exchange system by using Outlook Web App or Outlook Anywhere, and only authenticated users are allowed based on specified policies. Internal users can access the Internet through the TMG module. In the solution, two RH2285 V2 servers are configured in load balancing mode and the software is Microsoft's Forefront TMG. 14

23 One manager server MBS Structure The manager server is responsible for system protection. System protection involves protecting the operating system against viruses and ensuring security of servers. In the solution, the mainstream SEP antivirus software is used. An SEP client is deployed for each server. The SEPM is installed on the manager server. The manager server manages all SEP clients on the LAN. That is, software deployment, policy configuration, and engine upgrade are completed on the manager server. MBS Structure for the Low-cost 600-user Solution Figure 2-7 shows the MBS structure for the low-cost 600-user solution. Figure 2-7 MBS structure for the low-cost 600-user solution MAPI network Heartbeat and replication network It is recommended that each DAG have two networks: a messaging application programming interface (MAPI) network and a replication network. This ensures network and network path redundancy. In addition, the system can distinguish server faults from network faults. The system cannot distinguish the two types of faults if a single network is used. In the solution, the service network (MAPI network) and the replication network (heartbeat network) use independent network interface cards (NICs) for isolation purposes. On each server, the first two disks are configured as a Redundant Array of Independent Disks 1 (RAID 1) group for installing the operating system. The other six disks are reserved for data storage. The two servers form a DAG. On both servers, each two disks form a RAID 1 group for accommodating mailbox databases and logs. Each mailbox database has an active copy and a passive copy, ensuring HA. MBS Structure for the Low-cost 1200-user Solution Figure 2-8 shows the MBS structure for the low-cost 1200-user solution. 15

24 Figure 2-8 MBS structure for the low-cost 1200-user solution MAPI network Heartbeat and replication network It is recommended that each DAG have two networks: an MAPI network and a replication network. This ensures network and network path redundancy. In addition, the system can distinguish server faults from network faults. The system cannot distinguish the two types of faults if a single network is used. In the solution, the service network (MAPI network) and the replication network (heartbeat network) use independent NICs for isolation purposes. On each server, the last two disks are configured as a RAID 1 group for installing the operating system. The other 12 disks are reserved for data storage. The two servers form a DAG. On both servers, each two disks form a RAID 1 group for accommodating mailbox databases and logs. Each mailbox database has an active copy and a passive copy, ensuring HA. 16

25 MBS Structure for the Low-cost 1200-user Solution Figure 2-9 shows the MBS structure for the low-cost 1200-user solution. Figure 2-9 MBS structure for the low-cost 1200-user solution MAPI network Heartbeat and replication network It is recommended that each DAG have two networks: an MAPI network and a replication network. This ensures network and network path redundancy. In addition, the system can distinguish server faults from network faults. The system cannot distinguish the two types of faults if a single network is used. In the solution, the service network (MAPI network) and the replication network (heartbeat network) use independent NICs for isolation purposes. On each server, the first two disks are configured as a RAID 1 group for installing the operating system. Two SSD cards are installed on each server for data storage. The two servers form a DAG. Each SSD card is a partition for accommodating mailbox databases and logs. Each mailbox database has an active copy and a passive copy, ensuring HA. 17

26 MBS Structure for the Low-cost 3000-user Solution Figure 2-10 shows the MBS structure for the low-cost 3000-user solution. Figure 2-10 MBS structure for the low-cost 3000-user solution MAPI network Heartbeat and replication network It is recommended that each DAG have two networks: an MAPI network and a replication network. This ensures network and network path redundancy. In addition, the system can distinguish server faults from network faults. The system cannot distinguish the two types of faults if a single network is used. In the solution, the service network (MAPI network) and the replication network (heartbeat network) use independent NICs for isolation purposes. On each server, the last two disks are configured as a RAID 1 group for installing the operating system. The other 12 disks are reserved for data storage. The three servers form a DAG. Each mailbox database has an active copy and two passive copies, ensuring HA. 2.4 Server Performance Design The server performance design involves planning the central processing unit (CPU) computing capability and memory for the MBS, HTS, and CAS. 18

27 2.4.1 Server Performance Design for the Low-cost 600-user Solution CPU Configuration The CPU configuration requirements for the 600-user solution are described in the following aspects: Number of mailboxes: 600. Mailbox profile: A mailbox receives and sends 200 mails on average per day, with the average mail size 75 KB. Availability: A mailbox resiliency mechanism is available. The system can survive double-server failure events. Storage system: Each database has two copies and hosts 300 mailboxes. There are a total of four database copies residing on two nodes. Activation model: In the solution, a server is capable of hosting all active mailboxes when critical faults occur. That is, the system can survive single-server failure events, with the shortest interruption time. That is, when a server is faulty, the two database copies on the other server are activated. Server platform: one quad-core Intel Xeon E GHz CPU. Table 2-6 lists the CPU consumption statistics. Table 2-6 CPU consumption statistics Estimated CPU Consumption per Mailbox Number of Mails Received/Sent per Mailbox per Day (Average Mail Size75 KB) Megacycles for Active Mailbox Megacycles for Passive Mailbox The CPU consumption statistics are sourced from the help document of Microsoft Exchange Server The unit megacycles used in the reference system HP DL380 G5 x5470 is applied as the consumption unit. The reference CPU configuration is used, that is, two quad-core Intel Xeon x GHz CPUs are configured and each CPU core generates

28 megacycles. The value of SPECint_rate2006 of an eight-core CPU is 150 in the reference system HP DL380 G5 x GHz. That is, the value of each core is 18.75, which is referred to as the baseline value of each core. The number of megacycles of servers in the solution is calculated as follows: Number of megacycles of each core after adjustment = (New platform of each core value x Hz of each core on the reference platform)/reference value of each core In the solution, the RH2285 V2 configured with a quad-core Intel Xeon E GHz CPU is used. The value of SPECint_rate2006 is 106, that is, the value of each core is 26. The number of megacycles of each core is 4621, that is, the number of megacycles of a server is The number of megacycles of each core is calculated as follows: 26 x 3333/18.75 = 4621 I. CPU Design for the MBS Perform the following steps: Step 1 Calculate the number of MBSs. A DAG with two nodes is required so that the system can survive single-server failure events. Therefore, two MBSs are required. Step 2 Calculate the maximum number of active mailboxes hosted by each MBS based on the activation model. Assume that active database copies are evenly distributed on the MBSs. Each MBS hosts 300 active mailboxes in ideal conditions. In this solution, when an MBS is faulty, the other MBS must host 600 active mailboxes. Set MaximumActiveDatabase in Set-MailboxServer cmdlet to 2. Step 3 Calculate the estimated CPU requirement of the active MBS. According to Table 2-1, a mailbox receives and sends 200 mails on average per day, with the average mail size 75 KB. Each mailbox consumes 4 MHz megacycles. An MBS hosts a maximum of 600 live mailboxes. Therefore, the number of megacycles of an MBS is 2400 MHz. 600 x 4 MHz = 2400 MHz Step 4 Calculate the estimated CPU requirement of the passive MBS. Actually no passive MBS exists after the active MBS becomes faulty. Step 5 Calculate the estimated total CPU requirement. In the solution, the total estimated CPU requirement refers to the CPU requirement of the active MBS. Step 6 Apply the estimated total CPU requirement to the hardware platform. In the solution, the RH2285 V2 configured with a quad-core Intel Xeon E GHz CPU is used. The number of megacycles of each core is 4621, that is, the number of megacycles of a server is You are advised to set the maximum CPU usage of a server in peak hours to 70% when no fault occurs, and to 80% when faults occur. To sum up, the estimated total CPU requirement is 3000 (2400/0.8) MHz. That is, each MBS needs to be configured with a single E processor core. In practice, two E processor cores are configured for redundancy purposes. 20

29 The solution meets the CPU requirements of the MBSs. ----End II. CPU Design for the Composite HTS/CAS A balance must be achieved between the composite HTS/CAS and the MBS in resource consumption; otherwise, system performance is affected. Table 2-7 lists the mapping between server roles and recommended processor core ratios. Table 2-7 Mapping between server roles and recommended processor core ratios Server Role MBS:Global catalog MBS:HTS Recommended Processor Core Ratio 8:1 (64-bit global catalog) 4:1 (32-bit global catalog) 7:1 (the HTS does not provide the virus scan function) 5:1 (the HTS provides the virus scan function) MBS:CAS 4:3 MBS:Composite CAS/HTS 1:1 The information highlighted in bold is referenced in the solution design. According to Table 2-7, the ratio of HTS/CAS CPU cores to MBS CPU cores is 1:1. Two E processor cores needs to be configured for the HTS/CAS in the solution. In the solution, the MBS, HTS, and CAS are deployed on the same server. The estimated CPU usage in peak hours (when SPOF occurs) is calculated as follows: CPU usage in peak hours = Megacycles required by the three server roles/available megacycles of the server platform 3000 x 2/18484 = 32%. Memory Configuration III. CPU Design for the ETS According to Table 2-7, the ratio of HTS CPU cores to MBS CPU cores is 1:5. Therefore, two E cores are enough. A quad-core Intel Xeon E CPU is configured for each ETS because third-party antivirus software may be installed and antispam operations may be performed. Memory configuration involves configuring memory for the MBS, HTS, and CAS. Set the memory size of the MBS to the larger value of the size of memory consumed by live mailboxes and the minimum supported memory size. That is, if the consumed memory size is larger than the minimum supported memory size, set the memory size of the MBS to the consumed memory size. The HTS and CAS are installed on the same server. Therefore, you are advised to configure 2 GB of memory for each core for the composite HTS/CAS, and configure 1 GB for each core for the HTS. Table 2-8 lists the memory configurations recommended by Microsoft. 21

30 Table 2-8 Memory configurations recommended by Microsoft Server Role Minimum Configuration Recommended Value ETS 4 GB 1 GB for each core (totally 4 GB) HTS 4 GB 1 GB for each core (totally 4 GB) CAS 4 GB 2 GB for each core (totally 8 GB) Unified message server (UMS) 4 GB 2 GB for each core (totally 4 GB) MBS 4 GB 4 GB memory plus 330 MB additional space for each mailbox The required memory size depends on the mailbox profile and size of the database cache. Composite HTS/CAS (the HTS and CAS run on the same server) Multi-role server (the CAS, HTS, and MBS run on the same server) 4 GB 2 GB for each core (totally 8 GB) 8 GB 4 GB memory plus 330 MB additional space for each mailbox The required memory size depends on the mailbox profile and size of the database cache. According to Table 2-8, the multi-role server must be configured with at least 8 GB of memory. It is recommended that each MBS be configured with 4 GB plus 3-30 MB(The required memory size depends on the mailbox profile and size of the database cache.) additional space. The information highlighted in bold is referenced in the solution design. I. Memory Design for the MBS According to Table 2-5, if the 200 mail per day profile is used, each mailbox requires 12 MB of database cache and each MBS hosts a maximum of 600 live mailboxes (when SPOF occurs). The required memory size is calculated as follows: Minimum required database cache = MAX [(4 GB + Number of live mailboxes x Required memory size/1024), Minimum memory size for databases] = MAX [(4 GB x 12/1024 GB), 8 GB] = MAX [11 GB, 8 GB] = 11 GB If the multi-role architecture is used, a total of 11 GB of memory is required for the solution. It is recommended that 24 GB be configured. 22

31 Table 2-9 Default mailbox database cache sizes Server Physical Memory Database Cache Size (Mailbox Role Only) Database Cache Size (Multi-role) 2 GB 512 MB Not supported 4 GB 1 GB Not supported 8 GB 3.6 GB 2 GB 16 GB 10.6 GB 8 GB 24 GB 17.6 GB 14 GB 32 GB 24.4 GB 20 GB 48 GB 39.2 GB 32 GB 64 GB 53.6 GB 44 GB 96 GB 82.4 GB 68 GB 128 GB GB 92 GB The information highlighted in bold is referenced in the solution design. II. Memory Design for the ETS According to Table 2-8, the ETS is installed on an independent server. Therefore, you are advised to configure 1 GB of memory for each core. A total of 4 GB of memory is required. It is recommended that 8 GB be configured for the ETS Server Performance Design for the Low-cost 1200-user Solution CPU Configuration The CPU configuration requirements for the 1200-user solution are described in the following aspects: Number of mailboxes: Mailbox profile: A mailbox receives and sends 200 mails on average per day, with the average mail size 75 KB. Availability: A mailbox resiliency mechanism is available. The system can survive double-server failure events. Storage system: Each database has two copies and hosts 300 mailboxes. There are a total of eight database copies residing on two nodes. Activation model: In the solution, a server is capable of hosting all active mailboxes when critical faults occur. That is, the system can survive single-server failure events, with the shortest interruption time. That is, when a server is faulty, the four database copies on other servers are activated. Server platform: two quad-core Intel Xeon E GHz CPUs. 23

32 In the solution, the RH2285 V2 configured with two quad-core Intel Xeon E GHz CPUs is used. The value of SPECint_rate2006 is 208, that is, the value of each core is 26. The number of megacycles of each core is 4621, that is, the number of megacycles of a server is The number of megacycles of each core is calculated as follows: 26 x 3333/18.75 = 4621 I. CPU Design for the MBS Perform the following steps: Step 1 Calculate the number of MBSs. A DAG with two nodes is required so that the system can survive single-server failure events. Therefore, two MBSs are required. Step 2 Calculate the maximum number of active mailboxes hosted by each MBS based on the activation model. Assume that active database copies are evenly distributed on the MBSs. Each MBS hosts 600 active mailboxes in ideal conditions. In this solution, when an MBS is faulty, the other MBS must host 1200 active mailboxes. Set MaximumActiveDatabase in Set-MailboxServer cmdlet to 4. Step 3 Calculate the estimated CPU requirement of the active MBS. According to Table 2-2, a mailbox receives and sends 200 mails on average per day, with the average mail size 75 KB. Each mailbox consumes 4 MHz megacycles. An MBS hosts a maximum of 1200 live mailboxes. Therefore, the number of megacycles of an MBS is 4800 MHz x 4 MHz = 4800 MHz Step 4 Calculate the estimated CPU requirement of the passive MBS. Actually no passive MBS exists after the active MBS becomes faulty. Step 5 Calculate the estimated total CPU requirement. In the solution, the total estimated CPU requirement refers to the CPU requirement of the active MBS. Step 6 Apply the estimated total CPU requirement to the hardware platform. In the solution, the RH2285 V2 configured with two quad-core Intel Xeon E GHz CPUs is used. The number of megacycles of each core is 4621, that is, the number of megacycles of a server is You are advised to set the maximum CPU usage of a server in peak hours to 70% when no fault occurs, and to 80% when faults occur. To sum up, the estimated total CPU requirement is 6000 (4800/0.8) MHz. That is, each MBS needs to be configured with two E processor cores. In practice, four E processor cores are configured for redundancy purposes. The solution meets the CPU requirements of the MBSs. ----End II. CPU Design for the Composite HTS/CAS A balance must be achieved between the composite HTS/CAS and the MBS in resource consumption; otherwise, system performance is affected. 24

33 According to Table 2-7, the ratio of HTS/CAS CPU cores to MBS CPU cores is 1:1. Only one quad-core Intel Xeon E CPU needs to be configured for the HTS/CAS in the solution. In the solution, the MBS, HTS, and CAS are deployed on the same server. The estimated CPU usage in peak hours (when SPOF occurs) is calculated as follows: CPU usage in peak hours = Megacycles required by the three server roles/available megacycles of the server platform 4800 x 2/36,968 = 26% Memory Configuration III. CPU Design for the ETS According to Table 2-8, the ratio of ETS CPU cores to MBS CPU cores is 1:5. Therefore, two E cores are enough. A quad-core Intel Xeon E CPU is configured for each ETS because third-party antivirus software may be installed and antispam operations may be performed. Memory configuration involves configuring memory for the MBS, HTS, and CAS. Set the memory size of the MBS to the larger value of the size of memory consumed by live mailboxes and the minimum supported memory size. That is, if the consumed memory size is larger than the minimum supported memory size, set the memory size of the MBS to the consumed memory size. The HTS and CAS are installed on the same server. Therefore, you are advised to configure 2 GB of memory for each core for the composite HTS/CAS, and configure 1 GB for each core for the HTS. Table 2-8 lists the memory configurations recommended by Microsoft. I. Memory Design for the MBS, HTS, and CAS According to Table 2-5, if the 200 mail per day profile is used, each mailbox requires 12 MB of database cache and each MBS hosts a maximum of 1200 live mailboxes (when SPOF occurs). The required memory size is calculated as follows: Minimum required database cache = MAX [(4 GB + Number of live mailboxes x Required memory size/1024), Minimum memory size for databases] = MAX [(4 GB x 12/1024 GB), 8 GB] = MAX [18 GB, 8 GB] = 18 GB If the multi-role architecture is used, a total of 18 GB of memory is required for the solution. It is recommended that 32 GB be configured. Table 2-10 Default mailbox database cache sizes Server Physical Memory Database Cache Size (Mailbox Role Only) Database Cache Size (Multi-role) 2 GB 512 MB Not supported 4 GB 1 GB Not supported 8 GB 3.6 GB 2 GB 25

34 16 GB 10.6 GB 8 GB 24 GB 17.6 GB 14 GB 32 GB 24.4 GB 20 GB 48 GB 39.2 GB 32 GB 64 GB 53.6 GB 44 GB 96 GB 82.4 GB 68 GB 128 GB GB 92 GB The information highlighted in bold is referenced in the solution design. II. Memory Design for the ETS According to Table 2-8, the ETS is installed on an independent server. Therefore, you are advised to configure 1 GB of memory for each core. A total of 4 GB of memory is required. It is recommended that 8 GB be configured for the ETS Server Performance Design for the High-performance 1200-user Solution CPU Configuration The CPU configuration requirements for the 1200-user solution are described in the following aspects: Number of mailboxes: Mailbox profile: A mailbox receives and sends 500 mails on average per day, with the average mail size 75 KB. Availability: A mailbox resiliency mechanism is available. The system can survive single-server failure events. Storage system: Each database has two copies and hosts 600 mailboxes. There are a total of four database copies residing on two nodes. Activation model: In the solution, a server is capable of hosting all active mailboxes when critical faults occur. That is, the system can survive single-server failure events, with the shortest interruption time. That is, when a server is faulty, the two database copies on the other server are activated. Server platform: two quad-core Intel Xeon E GHz CPUs. Table 2-11 lists the CPU consumption statistics. 26

35 Table 2-11 CPU consumption statistics Estimated CPU Consumption per Mailbox Number of Mails Received/Sent per Mailbox per Day (Average Mail Size75 KB) Megacycles for Active Mailbox Megacycles for Passive Mailbox The CPU consumption statistics are sourced from the help document of Microsoft Exchange Server The unit megacycles used in the reference system HP DL380 G5 x5470 is applied as the consumption unit. The reference CPU configuration is used, that is, two quad-core Intel Xeon x GHz CPUs are configured and each processor core generates 3333 megacycles. The value of SPECint_rate2006 of an eight-core CPU is 150 in the reference system HP DL380 G5 x GHz. That is, the value of each core is 18.75, which is referred to as the baseline value of each core. The number of megacycles of servers in the solution is calculated as follows: Number of megacycles of each core after adjustment = (New platform of each core value x Hz of each core on the reference platform)/reference value of each core In the solution, the RH2285 V2 configured with two quad-core Intel Xeon E GHz CPUs is used. The value of SPECint_rate2006 is 208, that is, the value of each core is 26. The number of megacycles of each core is 4621, that is, the number of megacycles of a server is The number of megacycles of each core is calculated as follows: 26 x 3333/18.75 = 4621 I. CPU Design for the MBS Perform the following steps: Step 1 Calculate the number of MBSs. A DAG with two nodes is required so that the system can survive single-server failure events. Therefore, two MBSs are required. 27

36 Step 2 Calculate the maximum number of active mailboxes hosted by each MBS based on the activation model. Assume that active database copies are evenly distributed on the MBSs. Each MBS hosts 600 active mailboxes in ideal conditions. In this solution, when an MBS is faulty, the other MBS must host 1200 active mailboxes. Set MaximumActiveDatabase in Set-MailboxServer cmdlet to 4. Step 3 Calculate the estimated CPU requirement of the active MBS. According to Table 2-3, a mailbox receives and sends 500 mails on average per day, with the average mail size 75 KB. Each mailbox consumes 10 MHz megacycles. An MBS hosts a maximum of 1200 live mailboxes. Therefore, the number of megacycles of an MBS is 12,000 MHz. Step 4 Calculate the estimated CPU requirement of the passive MBS. Actually no passive MBS exists after the active MBS becomes faulty. Step 5 Calculate the estimated total CPU requirement. In the solution, the total estimated CPU requirement refers to the CPU requirement of the active MBS. Step 6 Apply the estimated total CPU requirement to the hardware platform. In the solution, the RH2285 V2 configured with two quad-core Intel Xeon E GHz CPUs is used. The number of megacycles of each core is 4621, that is, the number of megacycles of a server is You are advised to set the maximum CPU usage of a server in peak hours to 70% when no fault occurs, and to 80% when faults occur. To sum up, the estimated total CPU requirement is 15,000 (12,000/0.8) MHz. That is, each MBS needs to be configured with four E processor cores. The solution meets the CPU requirements of the MBSs. ----End II. CPU Design for the Composite HTS/CAS A balance must be achieved between the composite HTS/CAS and the MBS in resource consumption; otherwise, system performance is affected. According to Table 2-7, the ratio of HTS/CAS CPU cores to MBS CPU cores is 1:1. Only one quad-core Intel Xeon E CPU needs to be configured for the HTS/CAS in the solution. In the solution, the MBS, HTS, and CAS are deployed on the same server. The estimated CPU usage in peak hours (when SPOF occurs) is calculated as follows: CPU usage in peak hours = Megacycles required by the three server roles/available megacycles of the server platform 12,000 x 2/36,968 = 65% III. CPU Design for the ETS According to Table 2-8, the ratio of ETS CPU cores to MBS CPU cores is 1:5. Therefore, two E cores are enough. A quad-core Intel Xeon E CPU is configured for each ETS because third-party antivirus software may be installed and antispam operations may be performed. 28

37 Memory Configuration Memory configuration involves configuring memory for the MBS, HTS, and CAS. Set the memory size of the MBS to the larger value of the size of memory consumed by live mailboxes and the minimum supported memory size. That is, if the consumed memory size is larger than the minimum supported memory size, set the memory size of the MBS to the consumed memory size. The HTS and CAS are installed on the same server. Therefore, you are advised to configure 2 GB of memory for each core for the composite HTS/CAS, and configure 1 GB for each core for the HTS. Table 2-8 lists the memory configurations recommended by Microsoft. I. Memory Design for the MBS, HTS, and CAS According to Table 2-5, if the 500 mails per day profile is used, each mailbox requires 30 MB of database cache and each MBS hosts a maximum of 1200 live mailboxes (when SPOF occurs). The required memory size is calculated as follows: Minimum required database cache = MAX [(4 GB + Number of live mailboxes x Required memory size/1024), Minimum memory size for databases] = MAX [(4 GB x 30/1024 GB), 8 GB] = MAX [39.2 GB, 8 GB] = 39.2 GB If the multi-role architecture is used, a total of 64 GB of memory is required for the solution. Table 2-12 Default mailbox database cache sizes Server Physical Memory Database Cache Size (Mailbox Role Only) Database Cache Size (Multi-role) 2 GB 512 MB Not supported 4 GB 1 GB Not supported 8 GB 3.6 GB 2 GB 16 GB 10.6 GB 8 GB 24 GB 17.6 GB 14 GB 32 GB 24.4 GB 20 GB 48 GB 39.2 GB 32 GB 64 GB 53.6 GB 44 GB 96 GB 82.4 GB 68 GB 128 GB GB 92 GB The information highlighted in bold is referenced in the solution design. 29

38 II. Memory Design for the ETS According to Table 2-8, the ETS is installed on an independent server. Therefore, you are advised to configure 1 GB of memory for each core. A total of 4 GB of memory is required. It is recommended that 8 GB be configured for the ETS Server Performance Design for the Low-cost 3000-user Solution CPU Configuration The CPU configuration requirements for the 3000-user solution are described in the following aspects: Number of mailboxes: Mailbox profile: A mailbox receives and sends 200 mails on average per day, with the average mail size 75 KB. Availability: A mailbox resiliency mechanism is available. The system can survive double-server failure events. Storage system: There are three copies for each database. Each database contains 300 mailboxes and there are a total of 30 database copies residing on three nodes. Activation model: In the solution, when two servers are faulty, the other server carries all mail services. That is, when two servers are faulty, all 10 database copies of the other server are activated. Server platform: two six-core Intel Xeon E GHz CPUs. In the solution, the RH2285 V2 configured with two six-core Intel Xeon E GHz CPUs is used. The value of SPECint_rate2006 is 410, that is, the value of each core is The number of megacycles of each core is 6074, that is, the number of megacycles of a server is The number of megacycles of each core is calculated as follows: x 3333/18.75 = 6074 I. CPU Design for the MBS Perform the following steps: Step 1 Calculate the number of MBSs. A DAG with three nodes is required so that the system can survive double-server failure events. Therefore, three MBSs are required. Step 2 Calculate the maximum number of active mailboxes hosted by each MBS based on the activation model. Assume that active database copies are evenly distributed on the MBSs. Each MBS hosts 1000 active mailboxes in ideal conditions. In this solution, when an MBS is faulty, the other MBS must host 3000 active mailboxes. Set MaximumActiveDatabase in Set-MailboxServer cmdlet to 10. Step 3 Calculate the estimated CPU requirement of the active MBS. According to Table 2-4, a mailbox receives and sends 200 mails on average per day, with the average mail size 75 KB. Each mailbox consumes 4 MHz megacycles. An MBS hosts a maximum of 3000 live mailboxes. Therefore, the number of megacycles of an MBS is 12,000 MHz. 30

39 Step 4 Calculate the estimated CPU requirement of the passive MBS. Actually no passive MBS exists after two MBSs are faulty. Step 5 Calculate the estimated total CPU requirement. In the solution, the total estimated CPU requirement refers to the CPU requirement of the active MBS. Step 6 Apply the estimated total CPU requirement to the hardware platform. In the solution, the RH2285 V2 configured with two six-core Intel Xeon E GHz CPUs is used. The number of megacycles of each core is 6074, that is, the number of megacycles of a server is You are advised to set the maximum CPU usage of a server in peak hours to 70% when no fault occurs, and to 80% when faults occur. To sum up, the estimated total CPU requirement is 15,000 (12,000/0.8) MHz. That is, each MBS needs to be configured with three E processor cores. In practice, six E processor cores are configured for redundancy purposes. The solution meets the CPU requirements of the MBSs. ----End Memory Configuration II. CPU Design for the Composite HTS/CAS A balance must be achieved between the composite HTS/CAS and the MBS in resource consumption; otherwise, system performance is affected. According to Table 2-8, the ratio of HTS/CAS CPU cores to MBS CPU cores is 1:1. Only one six-core Intel Xeon E CPU needs to be configured for the HTS/CAS in the solution. In the solution, the MBS, HTS, and CAS are deployed on the same server. The estimated CPU usage in peak hours (when two nodes are faulty) is calculated as follows: CPU usage in peak hours = Megacycles required by the three server roles/available megacycles of the server platform 12,000 x 2/72,888 = 32% III. CPU Design for the ETS According to Table 2-8, the ratio of ETS CPU cores to MBS CPU cores is 1:5. Therefore, three E cores are enough. A six-core Intel Xeon E CPU is configured for each ETS because third-party antivirus software may be installed and antispam operations may be performed. Memory configuration involves configuring memory for the MBS, HTS, and CAS. Set the memory size of the MBS to the larger value of the size of memory consumed by live mailboxes and the minimum supported memory size. That is, if the consumed memory size is larger than the minimum supported memory size, set the memory size of the MBS to the consumed memory size. The HTS and CAS are installed on the same server. Therefore, you are advised to configure 2 GB of memory for each core for the composite HTS/CAS, and configure 1 GB for each core for the HTS. Table 2-8 lists the memory configurations recommended by Microsoft. 31

40 I. Memory Design for the MBS, HTS, and CAS According to Table 2-5, if the 200 mail per day profile is used, each mailbox requires 12 MB of database cache and each MBS hosts a maximum of 3000 live mailboxes (when SPOF occurs). The required memory size is calculated as follows: Minimum required database cache = MAX [(4 GB + Number of live mailboxes x Required memory size/1024), Minimum memory size for databases] = MAX [(4 GB x 12/1024 GB), 8 GB] = MAX [40 GB, 8 GB] = 40 GB If the multi-role architecture is used, a total of 40 GB of memory is required for the solution. It is recommended that 48 GB be configured. Table 2-13 Default mailbox database cache sizes Server Physical Memory Database Cache Size (Mailbox Role Only) Database Cache Size (Multi-role) 2 GB 512 MB Not supported 4 GB 1 GB Not supported 8 GB 3.6 GB 2 GB 16 GB 10.6 GB 8 GB 24 GB 17.6 GB 14 GB 32 GB 24.4 GB 20 GB 48 GB 39.2 GB 32 GB 64 GB 53.6 GB 44 GB 96 GB 82.4 GB 68 GB 128 GB GB 92 GB The information highlighted in bold is referenced in the solution design. II. Memory Design for the ETS According to Table 2-8, the ETS is installed on an independent server. Therefore, you are advised to configure 1 GB of memory for each core. A total of 6 GB of memory is required. It is recommended that 8 GB be configured for the ETS. 2.5 MBS Storage Design In HUAWEI Enterprise Solution, the hard disks of MBSs store data and no disk array needs to be deployed. This reduces the costs. 32

41 2.5.1 MBS Storage Design for the Low-cost 600-user Solution Storage Capacity Design Process In the 600-user solution, two RH2285 V2 servers are deployed and configured as a DAG for hosting database copies. Figure 2-3 shows the architecture of the solution. The architecture is described as follows: 1 TB 7.2K RPM 2.5-inch SATA disks are used for accommodating mailbox databases and transaction logs. On each server, four disks accommodate mailbox databases and transaction logs. The four disks are configured as two RAID 1 groups mapping two logical unit numbers (LUNs). Each LUN accommodates a mailbox database and its transaction logs. Each database and its corresponding transaction logs are placed on the same LUN. Each LUN has at least 10% to 20% free space. There are two HA mailbox database copies, with no lagged database copies. A restore LUN is deployed for maintenance and restoration. According to the low-level design described in section "Low-level Design": There are a total of 600 mailboxes and each database hosts 300 mailboxes. Each mailbox receives and sends an average of 200 mails per day and the mailbox quota is 1.5 GB. These mailboxes are evenly hosted by two MBSs. A mailbox receives an average of 75 MB of mails every five-day work week, with an average mail size of 75 KB. Single item recovery is enabled with a 14-day deleted item retention window. I. Size of a Single Mailbox Mailbox size = Mailbox limit + Whitespace + Dumpster Whitespace = 200 mails per day x 75/1024 MB = 15 MB Dumpster size = 200 mails per day x 75/1024 MB x 14 + (1.5 GB x GB x 0.03) x 1024 = 270 MB For details about the mailbox limit, blank space, and dumpster, access Table 2-14 lists the example values for determining the actual mailbox size on a disk. Table 2-14 Actual mailbox size on a disk Mailbox Quota Dumpster Size Whitespace Actual Mailbox Size 1.5 GB 270 MB 15 MB 1.78 GB II. Actual Database Size Database size = Number of mailboxes x Mailbox size x Database overhead growth factor 33

42 = 300 x 1.78 GB x (1 + 20%) = 641 GB Table 2-15 lists the database capacity requirements. Table 2-15 Database capacity requirements Number of Mailboxes per Database Number of Databases Database Size GB For details about the database overhead growth factor referenced in the preceding formula, access III. Log Capacity A mailbox using the 100 mail per day profile generates 40 transaction logs per day on average. Therefore, each database generates 12,000 logs per day. One percent of the mailboxes are moved once a week (on Saturday). The solution employs the native data protection function of the Exchange Server 2010, and therefore the logs generated in three consecutive days can be stored without truncation. Table 2-16 Log storage capacity Logs Generated per Database Log File Size Daily Log Size Move Mailbox Size/Database Truncation Log Size Requirements MB 11.7 GB 6.4 GB (6 x 1.78 x 1.2/2) 35.1 GB (3 x 11.7) 41.5 GB ( ) For details about the transaction logs, move mailbox size, and truncation failure tolerance, access IV. Disk Space for 300 Mailboxes Each database and its transaction logs are placed on the same LUN. The required LUN size is calculated as follows: LUN size = Database size/(1 Percentage of available space) = (Database size + Transaction log size + Content index size)/(1 0.2) = (641 GB GB GB)/0.9 =830 GB The available space of two formatted 1 TB SATA disks (RAID 1) is about 931 GB. The calculated size 830 GB is less than 931 GB. This is enough for accommodating the databases and transaction logs of 300 mailboxes, 1.5 GB for each mailbox. 34

43 In the Exchange Server 2010, the content index percentage is 10%. For details, access V. Disk Space for 600 Mailboxes Two 1 TB SATA disks configured as a RAID 1 group can accommodate 300 mailboxes. Four 1 TB 7.2K 2.5-inch SATA disks can accommodate the databases and transaction logs of 600 mailboxes, 1.5 GB for each mailbox. I/O Design In the solution, each server is configured with four 1 TB 7.2K RPM SATA disks. On an MBS, every two SATA disks configured as a RAID 1 group must be capable of accommodating an active mailbox database and its transaction logs (300 live mailboxes). Therefore, ensure that the mailbox I/O amount of a single database does not exceed the random I/O amount supported by a RAID 1 group. Because the solution leverages the just a bunch of disks (JBOD) architecture, ensure that the number of mailboxes hosted by each database does not exceed the random I/O amount supported by a RAID 1 group. Disk controllers with cache are used in the solution. Two 1 TB 7.2K RPM SATA disks configured as a RAID 1 group generate 100 random IOPS. Factoring in a 20% overhead growth buffer, a RAID 1 group can actually handle a total of 80 random IOPS. If the 200 mail per day profile is used, the estimated consumed IOPS of a mailbox is Therefore, a RAID 1 group supports a maximum of 330 mailboxes if this profile is used. The capacity design determines that 300 mailboxes are supported, but the profile determines that 330 mailboxes are supported. Therefore this solution can be deployed on a RAID 1 group with two disks. I. Disk IOPS Verification The Jetstress approves that two RAID 1 groups with four 1 TB 7.2K RPM SATA disks generate a maximum of 200 random IOPS. Factoring in a 20% overhead growth buffer, two RAID 1 groups can handle a total of 160 random IOPS. In practice, only 144 IOPS is required for accommodating 600 live mailboxes. Storage Design for the RH2285 V2 It is recommended that an independent LUN be configured as the restore LUN, and a mailbox databases and its transaction logs be placed on the same LUN. The disk requirements of an RH2285 V2 functioning as an MBS are as follows: Two 7.2K RPM 1 TB SATA disks configured as a RAID 1 group for deploying the operating system Four disks configured as two RAID 1 groups for accommodating 2 database copies and corresponding transaction logs, one disk for configuring the restore LUN, and the other functioning as a hot spare disk Figure 2-7 shows the layout of the eight disks. 35

44 2.5.2 MBS Storage Design for the Low-cost 1200-user Solution Storage Capacity Design Process In the low-cost 1200-user solution, two RH2285 V2 servers are deployed and configured as a DAG for hosting database copies. Figure 2-4 shows the architecture of the solution. The architecture is described as follows: 2 TB 7.2K 3.5-inch SATA disks are used for accommodating mailbox databases and transaction logs. On each server, the first eight disks accommodate mailbox databases and transaction logs. The eight disks are configured as four RAID 1 groups mapping four LUNs. Each LUN accommodates a mailbox database and its transaction logs. Each database and its corresponding transaction logs are placed on the same LUN. Each LUN has at least 10% to 20% free space. There are two HA mailbox database copies, with no lagged database copies. A restore LUN is deployed for maintenance and restoration. The MBS, HTS, and CAS run on the same server. According to the low-level design described in section "Low-level Design": There are a total of 1200 mailboxes and each database hosts 300 mailboxes. Each mailbox receives and sends an average of 200 mails per day and the mailbox quota is 3 GB. These mailboxes are evenly hosted by two MBSs. A mailbox receives an average of 75 MB of mails every five-day work week, with an average mail size of 75 KB. Single item recovery is enabled with a 14-day deleted item retention window. I. Size of a Single Mailbox Mailbox size = Mailbox limit + Whitespace + Dumpster Whitespace = 200 mails per day x 75/1024 MB = 15 MB Dumpster size = 200 mails per day x 75/1024 MB x 14 + (3 GB x GB x 0.03) x 1024 = 334 MB For details about the mailbox limit, blank space, and dumpster, access Table 2-17 lists the example values for determining the actual mailbox size on a disk. Table 2-17 Actual mailbox size Mailbox Quota Dumpster Size Whitespace Actual Mailbox Size 3 GB 334 MB 15 MB 3.34 GB 36

45 II. Database Size Database size = Number of mailboxes x Mailbox size x Database overhead growth factor = 300 x 3.34 GB x (1 + 20%) = 1202 GB Table 2-18 Database capacity requirements Number of Mailboxes per Database Number of Databases Database Size GB For details about the database overhead growth factor referenced in the preceding formula, access III. Log Capacity A mailbox using the 100 mail per day profile generates 40 transaction logs per day on average. Therefore, each database generates 12,000 logs per day. One percent of the mailboxes are moved once a week (on Saturday). The solution employs the native data protection function of the Exchange Server 2010, and therefore the logs generated in three consecutive days can be stored without truncation. Table 2-19 Log storage capacity Logs Generated per Database Log File Size Daily Log Size Move Mailbox Size/Database Truncation Log Size Requirements 12,000 1 MB 11.7 GB 12 GB (12 x 3.34 x 1.2/4) 35.1 GB (3 x 11.7) 47.1 GB ( ) For details about the transaction logs, move mailbox size, and truncation failure tolerance, access IV. Disk Space for 300 Mailboxes Each database and its transaction logs are placed on the same LUN. The required LUN size is calculated as follows: LUN size = Database size/(1 Percentage of available space) = (Database size + Transaction log size + Content index size)/(1 0.1) = (1202 GB GB GB)/0.9 = 1522 GB The available space of a formatted 2 TB SATA disk is about 1853 GB. The calculated size is 1522 GB. This is enough for accommodating the databases and transaction logs of 300 mailboxes, 3 GB for each mailbox. 37

46 In the Exchange Server 2010, the content index percentage is 10%. For details, access V. Disk Space for 1200 Mailboxes Two 2 TB SATA disks configured as a RAID 1 group can accommodate 300 mailboxes. Eight 2 TB 7.2K 3.5-inch SATA disks can accommodate the databases and transaction logs of 1200 mailboxes, 3 GB for each mailbox. I/O Design In the solution, each server is configured with eight 2 TB 7.2K RPM SATA disks. On an MBS, every two SATA disks configured as a RAID 1 group must be capable of accommodating an active mailbox database and its transaction logs (300 live mailboxes). Therefore, ensure that the mailbox I/O amount of a single database does not exceed the random I/O amount supported by a RAID 1 group. Because the solution leverages the JBOD architecture, ensure that the number of mailboxes hosted by each database does not exceed the random I/O amount supported by a RAID 1 group. Disk controllers with cache are used in the solution. Two 2 TB 7.2K RPM SATA disks configured as a RAID 1 group generate 100 random IOPS. Factoring in a 20% overhead growth buffer, a RAID 1 group can actually handle a total of 80 random IOPS. If the 200 mail per day profile is used, the estimated consumed IOPS of a mailbox is Therefore, a RAID 1 group supports a maximum of 330 mailboxes if this profile is used. The capacity design determines that 300 mailboxes are supported, but the profile determines that 330 mailboxes are supported. This solution can be deployed on a RAID 1 group with two disks. I. Disk IOPS Verification The Jetstress approves that four RAID 1 groups with eight 2 TB 7.2K RPM SATA disks generate a maximum of 400 random IOPS. Factoring in a 20% in an overhead growth buffer, four RAID 1 groups can handle a total of 320 random IOPS. In practice, only 288 IOPS is required for accommodating 1200 live mailboxes. Storage Design for the RH2285 V2 It is recommended that an independent LUN be configured as the restore LUN, and a mailbox databases and its transaction logs be placed on the same LUN. The disk requirements of an RH2285 V2 functioning as an MBS are as follows: Two 10K RPM 300 GB SAS disks configured as a RAID 1 group for deploying the operating system SAS: Serial Attached SCSI Eight 7.2K RPM 2 TB SATA disks configured as four RAID 1 groups for accommodating four database copies and corresponding transaction logs, one disk for configuring the restore LUN, and the other functioning as a hot spare disk Figure 2-8 shows the layout of the 12 disks. 38

47 2.5.3 MBS Storage Design for the High-performance 1200-user Solution Storage Capacity Design Process In the high-performance 1200-user solution, two RH2285 V2 servers are deployed and configured as a DAG for hosting database copies. Figure 2-5 shows the architecture of the solution. The architecture is described as follows: Huawei SSD cards are used for accommodating mailbox databases and transaction logs. The 2 SSD cards map two LUNs. Each LUN accommodates a mailbox database and its transaction logs. Each database and its corresponding transaction logs are placed on the same LUN. Each LUN has at least 10% to 20% free space. There are two HA mailbox database copies, with no lagged database copies. A restore LUN is deployed for maintenance and restoration. The MBS, HTS, and CAS run on the same server. According to the low-level design described in section "Low-level Design": There are a total of 1200 mailboxes and each database hosts 600 mailboxes. Each mailbox receives and sends an average of 500 mails per day and the mailbox quota is 0.5 GB. These mailboxes are evenly hosted by two MBSs. A mailbox receives an average of 75 MB of mails every five-day work week, with an average mail size of 75 KB. Single item recovery is enabled and a 7-day retention window is configured for deleted mails. I. Size of a Single Mailbox Mailbox size = Mailbox limit + Whitespace + Dumpster Whitespace = 500 mails per day x 75/1024 MB = 37 MB Dumpster size = 500 mails per day x 75/1024 MB x 7 + (0.3 GB x GB x 0.03) x 1024 = 269 MB For details about the mailbox limit, blank space, and dumpster, access Table 2-20 lists the example values for determining the actual mailbox size on a disk. Table 2-20 Actual mailbox size Mailbox Quota Dumpster Size Whitespace Actual Mailbox Size 0.3 GB 269 MB 37 MB 0.6 GB 39

48 II. Actual Database Size Database size = Number of mailboxes x Mailbox size x Database overhead growth factor = 600 x 0.6 GB x (1 + 20%) = 432 GB Table 2-21 Database capacity requirements Number of Mailboxes per Database Number of Databases Database Size GB For details about the database overhead growth factor referenced in the preceding formula, access III. Log Capacity A mailbox using the 500 mail per day profile generates 100 transaction logs per day on average. Therefore, each database generates 60,000 logs per day. One percent of the mailboxes are moved once a week (on Saturday). The solution employs the native data protection function of the Exchange Server 2010, and therefore the logs generated in a single day can be stored without truncation. Table 2-22 Log storage capacity Logs Generated per Database Log File Size Daily Log Size Move Mailbox Size/Database Truncation Log Size Requirements 60,000 1 MB 59 GB 4.32 GB (12 x 0.6 x 1.2/2) 59 GB (1 x 59) 182 GB ( ) For details about the transaction logs, move mailbox size, and truncation failure tolerance, access IV. Disk Space for 600 Mailboxes Each database and its transaction logs are placed on the same LUN. The required LUN size is calculated as follows: LUN size = Database size/(1 Percentage of available space) = (Database size + Transaction log size + Content index size)/(1 0.1) = (432 GB + 59 GB GB)/0.9 = 594 GB The available space of a formatted 640 GB SSD card is about 600 GB. The calculated size is 594 GB. This is enough for accommodating the databases and transaction logs of 600 mailboxes, 0.3 GB for each mailbox. 40

49 In the Exchange Server 2010, the content index percentage is 10%. For details, access V. Total Disk Space for 1200 Mailboxes A 640 GB SSD card can accommodate 600 mailboxes. Two 640 GB SSD cards can accommodate the databases and transaction logs of 1200 mailboxes, 0.3 GB for each mailbox. I/O Design In the solution, each server is configured with two 640 GB SSD cards. On an MBS, each SSD card must accommodate an active mailbox database and its transaction logs (600 live mailboxes). Therefore, ensure that the mailbox I/O amount of a single database does not exceed the random I/O amount supported by an SSD card. An SSD card generates at least 1500 random IOPS. Factoring in a 20% overhead growth buffer, an SSD card can actually handle a total of 1200 random IOPS. If the 500 mail per day profile is used, the estimated consumed IOPS of a mailbox is 0.6. Therefore, an SSD card supports a maximum of 2000 mailboxes if this profile is used. The capacity design determines that 600 mailboxes are supported, but the profile determines that 2000 mailboxes are supported. This solution meets the I/O performance. I. Disk IOPS Verification The Jetstress approves that two 640 GB SSD cards generate a maximum of 6000 random IOPS. Factoring in a 20% overhead growth buffer, two SSD cards can handle a total of 4800 random IOPS. In practice, only 720 IOPS is required for accommodating 1200 live mailboxes. Storage Design for the RH2285 V2 It is recommended that an independent LUN be configured as the restore LUN, and a mailbox databases and its transaction logs be placed on the same LUN. The disk requirements of an RH2285 V2 functioning as an MBS are as follows: Two 7.2K RPM 1 TB SATA disks configured as a RAID 1 group for deploying the operating system One 7.2K RPM 1 TB SATA disk functioning as a hot spare disk One 7.2K RPM 1 TB SATA disk for configuring the restore LUN Figure 2-9 shows the layout of the four disks MBS Storage Design for the Low-cost 3000-user Solution Storage Capacity Design Process In the low-cost 3000-user solution, three RH2285 V2 servers are deployed and configured as a DAG for hosting database copies. Figure 2-6 shows the architecture of the solution. The architecture is described as follows: 2 TB 7.2K 3.5-inch SATA disks are used for accommodating mailbox databases and transaction logs. On each server, the first 10 disks accommodate mailbox databases and transaction logs. 41

50 The 10 disks map 10 LUNs. Each LUN accommodates a mailbox database and its transaction logs. Each database and its corresponding transaction logs are placed on the same LUN. Each LUN has at least 10% to 20% free space. There are three HA mailbox database copies, with no lagged database copies. A restore LUN is deployed for maintenance and restoration. The MBS, HTS, and CAS run on the same server. According to the low-level design described in section "Low-level Design": There are a total of 3000 mailboxes and each database hosts 300 mailboxes. Each mailbox receives and sends an average of 200 mails per day and the mailbox quota is 3 GB. These mailboxes are evenly hosted by three MBSs. A mailbox receives an average of 75 MB of mails every five-day work week, with an average mail size of 75 KB. Single item recovery is enabled and a 14-day retention window is configured for deleted mails. I. Size of a Single Mailbox Mailbox size = Mailbox limit + Whitespace + Dumpster Whitespace = 200 mails per day x 75/1024 MB = 15 MB Dumpster size = 200 mails per day x 75/1024 MB x 14 + (3 GB x GB x 0.03) x 1024 = 334 MB For details about the mailbox limit, blank space, and dumpster, access Table 2-23 lists the example values for determining the actual mailbox size on a disk. Table 2-23 Actual mailbox size Mailbox Quota Dumpster Size Whitespace Actual Mailbox Size 3 GB 334 MB 15 MB 3.34 GB II. Actual Database Size Database size = Number of mailboxes x Mailbox size x Database overhead growth factor = 300 x 3.34 GB x (1 + 20%) =1202 GB Table 2-24 Database capacity requirements Number of Mailboxes per Database Number of Databases Database Size GB 42

51 For details about the database overhead growth factor referenced in the preceding formula, access III. Log Capacity A mailbox using the 200 mail per day profile generates 40 transaction logs per day on average. Therefore, each database generates 12,000 logs per day. One percent of the mailboxes are moved once a week (on Saturday). The solution employs the native data protection function of the Exchange Server 2010, and therefore the logs generated in three consecutive days can be stored without truncation. Table 2-25 Log storage capacity Logs Generated per Database Log File Size Daily Log Size Move Mailbox Size/Database Truncation Log Size Requirements 12,000 1 MB 11.7 GB 12 GB (30 x 3.34 x 1.2/10) 35.1 GB (3 x 11.7) 47.1 GB ( ) For details about the transaction logs, move mailbox size, and truncation failure tolerance, access IV. Disk Space for 300 Mailboxes Each database and its transaction logs are placed on the same LUN. The required LUN size is calculated as follows: LUN size = Database size/(1 Percentage of available space) = (Database size + Transaction log size + Content index size)/(1 0.2) = (1202 GB GB GB)/0.9 = 1521 GB The available space of a formatted 2 TB SATA disk is about 1853 GB. The calculated size is 1521 GB. This is enough for accommodating the databases and transaction logs of 300 mailboxes, 3 GB for each mailbox. In the Exchange Server 2010, the content index percentage is 10%. For details, access V. Disk Space for 3000 Mailboxes A 2 TB SATA disk can accommodate 300 mailboxes. Ten 2 TB 7.2K 3.5-inch SATA disks can accommodate the databases and transaction logs of 3000 mailboxes, 3 GB for each mailbox. I/O Design In the solution, each server is configured with ten 2 TB 7.2K RPM SATA disks. On an MBS, each SATA disk must be capable of accommodating an active mailbox database and its 43

52 transaction logs (3000 live mailboxes). Therefore, ensure that the mailbox I/O amount of a single database does not exceed the random I/O amount supported by a disk. Because the solution leverages the JBOD architecture, ensure that the number of mailboxes hosted by each database does not exceed the random I/O amount supported by a single disk. Disk controllers with cache are used in the solution. A single 2 TB 7.2K RPM SATA disk generates 100 IOPS. Factoring in a 20% overhead growth buffer, a disk can actually handle a total of 80 random IOPS. If the 200 mail per day profile is used, the estimated consumed IOPS of a mailbox is Therefore, a disk supports a maximum of 330 mailboxes if this profile is used. The capacity design determines that 300 mailboxes are supported, but the profile determines that 330 mailboxes are supported. This solution can be deployed on a single disk. I. Disk IOPS Verification The Jetstress approves that ten 2 TB 7.2K RPM SATA disks generate a maximum of 1000 random IOPS. Factoring in a 20% overhead growth buffer, 10 disks can handle a total of 800 random IOPS. In practice, only 720 IOPS is required for accommodating 3000 live mailboxes. Storage Design for the RH2285 V2 It is recommended that an independent LUN be configured as the restore LUN, and a mailbox databases and its transaction logs be placed on the same LUN. The disk requirements of an RH2285 V2 functioning as an MBS are as follows: Two 10K RPM 300 GB SAS disks configured as a RAID 1 group for deploying the operating system 10 disks for accommodating 10 database copies and corresponding transaction logs, one for configuring the restore LUN, and one functioning as a host spare disk. Figure 2-10 shows the layout of the 14 disks. 2.6 AD Design The planning and deployment of HUAWEI Enterprise Solution is closely related to the AD & DNS service. In HUAWEI Enterprise Solution, the AD service is deployed on Microsoft Windows Server 2008 R2, which helps information technology professionals control the infrastructure. The AD service can be provided by one or multiple servers. It manages global accounts and resources and provides mandatory functions required by the MBS. The information is automatically synchronized among organizations. Requirements for deploying the AD service: Ensure that the functional level of forests and domains is Windows Server 2003 or later, and that of global catalogs is Windows Server 2003 SP2 or later. It is strongly recommended that the functional level of global catalogs be Windows Server 2008 R2. This is because the performance of a 64-bit global catalog is twice that of a 32-bit global catalog under the same configuration. In HUAWEI Enterprise Solution, two Windows Server 2008 R2 servers function as AD & DNS controllers on which the DNS service is enabled. The running of the AD and DNS services determine the status of the Exchange Server At least two servers must be 44

53 configured for deploying two AD & DNS controllers in active/standby mode. The functional level of the AD service is Windows Server 2008 R2 for higher performance. According to Table 2-7, a single core is enough for a global catalog. It is recommended that four cores be configured for a global catalog. 2.7 Antispam and Antivirus Design Antispam Design Antivirus Design Design Rules There are many spam mails on the Internet and users spend much time in disposing of those mails. However, this is not a problem with a Huawei Exchange system. In HUAWEI Enterprise Solution, mailboxes have the intelligent message filtering function to block spam mails. The content filtering technology CloudMark is introduced in Exchange Server The sample library is updated every day. The spam identification rate reaches up to 99%, according to West Coast Labs in a latest test in The Exchange Server 2010 uses the memory-based antivirus technology and scans mails in the memory. This does not affect the I/O performance of disks. In addition, the Exchange Server 2010 is embedded with multiple antivirus engines, which are automatically upgraded. The Exchange Server 2010 chooses a proper antivirus engine intelligently. The feature code update speed is 38 times faster than that of single-engine antivirus systems. This effectively protects Huawei Exchange system against viruses. The antispam and antivirus design rules are as follows: Antispam filter update The Exchange Server 2010 uses Microsoft Update to ensure that the antispam filter remains in the latest state. By default, the administrator must access Microsoft Update to download and install the latest content filter if manual update is used. Content filter update data is released every two weeks. Microsoft IP Reputation Service data and spam signature data are manually updated from Microsoft Update. Microsoft IP Reputation Service data and spam signature data can be obtained only by using Forefront Security for Exchange Server. Antivirus software deployment on the ETS and HTS Transmission-based antivirus software functions as a transmission agent or includes a transmission agent. Forefront Protection for Exchange Server is integrated into the Exchange Server 2010 to provide antivirus protection. The antivirus software must be deployed on the first perimeter of a mail system, that is, the SMTP gateway that external mails must pass before being forwarded to the mail system. In the Exchange Server 2010, the ETS is the first perimeter. In the Exchange Server 2010, all mails are forwarded to the HTS for routing. That is, the HTS routes all external mails sent to the Exchange system, mails transmitted within the Exchange system, and mails of which the sending and receiving mailboxes reside on the 45

54 same MBS. To better prevent viruses, you are advised to deploy antivirus software on the HTS Antispam and Antivirus Requirements The antispam and antivirus requirements are as follows: At least two ETSs are deployed to ensure HA. Forefront Security for Exchange Server is installed on both the ETS and HTS Networking Scheme and Design Figure 2-11 shows the networking scheme for HUAWEI Enterprise Solution. Figure 2-11 Networking scheme ETS Design In the Exchange Server 2010, the ETS is deployed on the perimeter network. The ETS aims at minimizing the attack scope and handling all mails bound for the Internet. The ETS functions as an SMTP relay and intelligent host in the Exchange system. A series of agents run on the ETS to ensure security. When the ETS handles mails, the agents scan the mails. The agents provide antivirus and antispam functions and controls the mail flows based on certain rules. Computers where the ETS is installed cannot access ADs. All configuration and recipient information is stored in an Active Directory Lightweight Directory Services (AD LDS) instance. To search recipients, the ETS requires data in the AD. The data is synchronized to the ETS by using EdgeSync. EdgeSync is a group of processes that run on a computer where the HTS is installed. It copies recipients and configurations from the AD to the AD LDS instance on the ETS. EdgeSync copies information required by the ETS to perform antispam configuration tasks, and connector configuration information required when end-to-end mail flow is enabled. EdgeSync synchronizes data as planned to ensure that the AD LDS instance contains the latest information. 46

55 In this design, multiple ETSs are installed on the perimeter network. This ensures redundancy and failover. In addition, a cloned script is used to ensure configuration consistency among the ETSs. Antispam and Antivirus Software Design Forefront Protection 2010 for Exchange (FPE) is used to effectively protect the Exchange Server 2010 against viruses, worms, spy software, and spam mails. FPE uses multiple industry-leading antivirus engines to scan viruses, improving the efficiency. FPE also provides a user-friendly console for the administrator to perform tasks easily, including configuring FPE options, configuring filters, and monitoring and viewing reports. FPE supports Powershell so that the administrator can operate FPE more conveniently. FPE protects the Exchange Server 2010 in three aspects: ETS, HTS, and MBS. That is, mails are scanned by the three server roles when being forwarded into or out of the Exchange system. All scanned mails are labeled to avoid repeated scan. For efficiency and security purposes, it is recommended that FPE be installed on the ETS and HTS to scan inbound and outbound mails. Software Update Scheme Design To maximize efficiency, you are advised to define viruses and upgrade engines for FPE on a local agent server. Engine and virus definitions, and worm list updates can be automatically downloaded from a Microsoft Hypertext Transfer Protocol (HTTP) server or an FPE-enabled Exchange server. After FPE update is downloaded, the engine (if enabled) becomes offline. After the update is automatically installed, the engine becomes online. Within this period, other engines keep scanning malicious software. Engines are upgraded when a later version is detected. When virus definitions are updated, the updates are added to engines. Virus definitions are updated frequently whereas engines are not. Virus definition and engine updates are detected at specified intervals. If no new file is available, no content is downloaded for engines within the specified interval. You are advised to upgrade all the five engines but enable two or three of them for scanning. When an engine is upgraded, it is offline and does not scan. If the five engines are upgraded, FPE can choose another available engine when an engine is upgraded. The sizes of the upgrade files of the engines are different, ranging from 80 MB to 200 MB. The size of an incremental upgrade file is about 10 MB. An agent server connects to Microsoft HTTP servers on the Internet. The agent server can cache data. Therefore, you are advised to configure all FPE servers to directly connect to Microsoft HTTP server. After a complete upgrade is performed for an engine, only incremental update needs to be performed later. This does not generate heavy traffic. 2.8 Load Balancing Scheme Design Hardware Load Balancing Scheme In the Exchange Server 2010, members of the CAS cluster must work in load balancing mode. Load balancing can be achieved using software and hardware. In the software-based load balancing scheme, NLB is used. In the hardware-based load balancing scheme, F5 is used. 47

56 Figure 2-12 CAS access flowchart The NLB scheme provides low performance and has restrictions. Therefore, the hardware scheme F5 is recommended. As shown in Figure 2-12, all clients access CASs through the F5 Local Traffic Manager (LTM). Multiple CASs are configured as a cluster for connecting clients. In addition to load balancing services, the LTM supports security access and performance expedition. Figure 2-12 shows the CAS access flowchart. An LTM can meet the performance requirements. In practice, two LTMs are configured to improve reliability. As shown in Figure 2-13, on the LTM, a virtual server Internet Protocol (IP) address is configured for the CAS cluster. The cluster is composed of three nodes. The LTM can connect to the server directly or connect to the server through a switch. Figure 2-13 shows the CAS access mode. Figure 2-13 CAS access mode The F5 LTM supports three networking modes: one-arm, two-arm, and npath. The npath mode has many restrictions in application. For more information, see related documents about F5. The one-arm and two-arm modes are described in the following sections. 48

57 Two-arm Networking In two-arm networking, two service network segments connect to the core switch, and the virtual server IP address and pool member IP addresses respectively reside on the two network segments. The virtual server IP address is on the external virtual local area network (VLAN) and the network segment x. The pool member IP addresses are on the internal VLAN and network segment x. Figure 2-14 shows two-arm networking. Figure 2-14 Two-arm networking Firewall 1 Firewall 2 Switch 1 Switch 2 Serial cable Layer-4 switch 1 Layer-4 switch 2 Server 1 Server 2 Server 3 Server 4 Figure 2-15 shows the data flows in two-arm networking. Figure 2-15 Data flows in two-arm networking The LTM translates the destination IP address of packets into the node IP address based on the load. The LTM translates the source IP address of packets into the virtual server IP address based on the load. The two-arm networking mode has the following features: The server gateway points to the floating IP address of the internal VLAN. External clients cannot directly connect to servers. 49

58 One-arm Networking Servers are privy to actual IP addresses of clients. Data flows can be easily identified. All inbound and outbound data flows must pass F5 LTM. The secure network address translation (SNAT) or virtual server function must be enabled if servers need to access external networks. In one-arm networking, only one service network segment connects to the core switch, and the virtual server IP address and pool member IP addresses reside on this network segment. The virtual server IP address and pool member IP addresses are on the same network segment, that is, x. Figure 2-16 shows one-arm networking. Figure 2-16 One-arm networking Firewall 1 Firewall 2 Serial cable Switch 1 Switch 2 Layer-4 switch 1 Layer-4 switch 2 Server 1 Server 2 Server 3 Server 4 The one-arm networking mode has the following characteristics: The server gateway points to a Layer-3 switch. External clients and servers can visit each other directly. Only data packets that require load balancing pass F5 LTM. When data packets pass F5 LTM, source address translation is performed. Servers cannot learn actual IP addresses of clients. After passing F5 LTM, the source and destination IP addresses of data packets change after translation. This causes inconvenience in fault identification. The one-arm networking mode is easy to deploy and does not require a switch whereas the two-arm networking focuses on security and requires VLAN division on the switch. Select a proper networking mode based on actual requirements. 50

59 2.9 Backup Scheme Design Overall Requirements for the Backup System The backup system must ensure that mails in the Exchange Server 2010 can be backed up and restored. The backup objects include mails on MBSs and the DAG. Requirements for the backup system are as follows: The backup system can meet the backup requirements under different conditions. Table 2-26 Backup requirements Mailbox Quantity Mail Quantity Mail Size (KB) Estimated Daily Data Amount (GB) Estimated Weekly Data Amount (GB) The backup system can perform incremental backup for mails within 34 hours per day. The backup system can perform full backup for mails per week. Mails of the latest month can be restored. Data of only active database copies on the active server in the DAG is backed up Networking for the Backup Scheme Figure 2-17 shows the networking for the backup scheme. Figure 2-17 Networking for the backup scheme 51

60 The mail backup module consists of the core backup server (NBU_Master) and Exchange servers (NBU_Client). The mail backup network must be isolated from the service network. Mails of Exchange servers are backed up to the NBU_Master when mailbox users are offline Backup Capacity Calculation Capacity design is an important part of backup scheme design. The following is an example. Assume that: Initial data amount (A): 8.6 GB Estimated daily incremental data amount (B): 8.6 GB Full backup cycle (C): 7 days Incremental backup cycle (D): 1 day Retention period (E): 30 days Full backup times (E/C): 4 times per week Incremental backup times within a full backup cycle(c/d) Full backup data amount within a retention period = Full backup times x Initial data amount + (Full backup times 1) x Estimated daily incremental data amount x Incremental backup times within a full backup cycle/2 = E/C x A + E/C x (E/C 1) x B x C/D/2 = AE/C + EB (E/C 1)/D/2 = 8.6 x 30/ x 8.6 x (30/7 1)/1/2 470 GB Incremental backup data amount within a backup cycle = Incremental backup times x Estimated daily incremental data amount = E/D x B = 30/1 x GB The total capacity required by the backup media is calculated as follows: AE/C + EB (E/C 1)/D/2 + E/D x B 730 GB Old backup data is not deleted before new backup data is generated. Therefore, the actually required storage capacity is more than twice the capacity required by the backup media. Four redundancy purposes, it is recommended that at least three times the capacity required by the backup media be configured, that is, more than 2190 GB Performance Design for the Backup Scheme The backup data amount, backup window, and physical performance of the backup device are considered in the performance design for the backup scheme. Assume that the backup data amount is A GB, the backup window is B hours, and the transmission rate is C Mbit/s. The performance can be ensured if the following conditions are met: A x 1024/C < B x 3600 The data access rate is sufficient. 52

61 2.10 Archiving Scheme Design Overall Requirements for the Archiving System Mails are automatically archived and managed. Both mails and logs are archived. Archiving policies can be easily configured to reduce space occupation on the active disk, thereby ensuring that archived mails meet verification requirements. Requirements for the archiving system are as follows: The archiving system can meet the archiving requirements under the conditions described in Table Table 2-27 Archiving requirements Mailbox Quantity Daily Mail Quantity Mail Size (KB) Estimated Daily Data Amount (GB) Estimated Monthly Data Amount (GB) The archiving system can archive mails within 34 hours per month. Mails are archived every month. The archived mails are retained for three years. Users can query archived mails Networking for the Archiving Scheme Figure 2-18 shows the networking for the archiving scheme. Symantec's archiving software Enterprise Vault is used. The database management software Structured Query Language (SQL) Server needs to be configured. 53

62 Figure 2-18 Networking for the archiving scheme As shown in Figure 2-18, the mail archiving module comprises the core archiving server, database server (SQL Server), and Quantum tape library. The NBU_Master is required if archived mails need to be migrated from the Enterprise Vault server to the tape library Archiving Capacity Calculation If the Exchange Server 2010 generates 258 GB data per month and the data is retained for three years, the capacity required by the storage media is calculated as follows: 258 GB x 12 x 3 = 9288 GB 2.11 Design for Forefront TMG Overall Requirements In HUAWEI Enterprise Solution, users that access Exchange mailboxes by using Outlook Web App and Outlook Anywhere users need to be authenticated on the Internet. In addition, internal users need to access the Internet through the TMG for software upgrade. To meet the preceding requirements, Forefront TMG servers are deployed in the DMZ Overall Scheme Design In HUAWEI Enterprise Solution, firewalls are deployed at edges of the topology to protect an enterprise network. The corporate network is an internal network and external users access the enterprise network over the Internet. TMG servers reside in the DMZ between the internal network and the external network. Multiple TMG servers are configured into an array for HA purposes. In addition, hardware LB is configured so that the TMG servers handle client requests evenly. 54

63 Clients on the external network access Huawei Exchange system by using Outlook Web App (OWA) over Secure Sockets Layer (SSL). Forefront TMG obtains client information through listening and forwards the information to Exchange servers. The Exchange servers verify the information and return mailbox information only to authenticated clients. Forefront TMG also supports Outlook Anywhere. In this way, external users can access services provided by the Exchange Server 2010 as internal users do. Outlook Anywhere access is implemented by using the Remote Procedure Call (RPC) over HTTP. For security purposes, the TMG array is added to the enterprise domain. Servers in the TMG array identify each other based on workgroup or domain information. Figure 2-19 shows the architecture of the Forefront TMG Figure 2-19 Architecture of the Forefront TMG Architecture Design and Network Topology Forefront TMG 2010 supports four network topologies. Edge firewall: In this topology, Forefront TMG is deployed at the network edge and functions as the organization's firewall. The TMGs connect to two networks: the internal network and the external network (usually the Internet). 3-Leg perimeter: This topology implements a perimeter network. Forefront TMG is connected to at least three physical networks: the internal network, one or more perimeter networks, and the external network. Back firewall: In this topology, Forefront TMG is located at the network's backend. Use this topology when another network element, such as a perimeter network or an edge security device, is located between Forefront TMG and the external network. Forefront TMG is connected to the internal network and to the network element in front of it. Single NIC: This topology enables limited Forefront TMG functionality. In this topology, Forefront TMG is connected to only one network, either the internal network or perimeter network. Typically, you must use this configuration when Forefront TMG is located on the internal corporate network or on a perimeter network, and another firewall is located at the edge, protecting corporate resources from the Internet. Forefront TMG can be connected to the LAN directly or connected to the LAN through a router or another firewall HA and Scalability Design Forefront TMG 2010 Enterprise that supports multiple-server arrays is used to ensure high performance and HA. You can configure Forefront TMG to distribute outbound traffic between two Internet service provider (ISP) connections by using one of the following methods: 55

64 Load balancing with failover capabilities: HA between two connections, including the following capabilities: Load balancing: Distribute traffic among the connections according to the ratio you define. For example, you can allocate 80% of traffic to one connection and the remaining 20% to the other connection. Failover: If one connection becomes unavailable, traffic is handled by the other connection. Internet connection is uninterrupted and end users are not affected. Failover only: One connection is defined as the primary connection for all traffic, whereas the other connection serves only as the backup connection. If the primary connection becomes unavailable, traffic is routed to the backup connection and Internet service is uninterrupted. HUAWEI Enterprise Solution uses load balancing with failover capabilities Publishing Design Forefront TMG enables you to make internal applications and services available to both internal and external users. When publishing Web servers, Forefront TMG uses Web publishing rules to allow or deny access to internal Web applications based on access policies. You can restrict access to specified users, computers, or networks, require user authentication, and check the traffic between clients and the publishing servers. HUAWEI Enterprise Solution uses the following Web publishing methods: Publishing Web servers over the Hypertext Transfer Protocol Secure (HTTPS): Publish a single website or load balancer, multiple websites, or a server farm over HTTPS. OWA publishing: OWA is an Exchange mail service that allows users to access their Exchange mailbox from a Web browser. Web listeners are used in HUAWEI Enterprise Solution. Each Forefront TMG Web publishing rule is assigned a Web listener. The Web listener listens on incoming connections on the defined networks or incoming connections using defined IP addresses and ports. The Web listener also defines the number of concurrent client connections that are allowed on the connection, and the authentication method that is used if authentication is required Web Access Design Array Design Forefront TMG provides Web access control and protection for internal users accessing the Internet by providing authentication, packet filtering, stateful inspection, and application layer filtering. Authenticating internal users Controlling Web access Inspecting and filtering Web traffic Accelerating Web access Forefront TMG has the following features: HA: ensuring uninterrupted running of Forefront TMG even during the downtime of one or more TMG servers. The TMG configurations of all servers in the array are identical to ensure uninterrupted service when one or more servers in the array fail. 56

65 LB Design Scalability: meeting increasing performance requirements. For example, additional network bandwidth is required for users that want to extend their Internet activities. To meet the organization's growing needs, you can upgrade a standalone Forefront TMG to a Forefront TMG array and increase the numbers of array members and arrays. Distributed permanent caching: enabling users to choose a new array manager. All servers are configured and upgraded by using the latest array manager. The information is permanent and reserved when one or more TMG servers fail. In Forefront TMG, when you publish a farm of Web servers that perform the same role or have the same content, you can enable HA for inbound access by configuring Forefront TMG to balance load among the servers in the farm. Load balancing ensures that requests are distributed evenly among the available Web servers, detects offline servers, implements failover, and maintains farm servers, without interrupting current endpoint connections Protection Design In this document, system protection refers to the software antivirus scheme on servers, which does not involve external hardware firewalls. System protection involves defense against viruses (worms and Trojan horse), spy software (Rootkit and advertisement software), proactive threats (SONAR), and network threats (firewall protection and intrusion detection). SONAR: Symantec Online Network for Advanced Response The design of the protection scheme involves deploying and managing antivirus software. There are two management schemes. Scheme 1: Unified Management The SEP software is installed on each server and a server with SEPM installed is deployed as the manager server. SEP is Symantec's antivirus software for operating systems and SEPM is the corresponding management software. You can install SEP, upgrade antivirus engines, and configure antivirus policies for each server on the SEPM console. The SEPM server can upgrade the antivirus engines by connecting to the Internet or using a Universal Serial Bus (USB) flash drive. The update is then synchronized to all servers. This scheme has two advantages. Security: The antivirus engines of all servers are upgraded by the SEPM console and therefore the servers do not need to connect to the Internet, thereby protecting the servers against viruses from the Internet. Convenience: The administrator performs unified deployment and management of antivirus software for all servers on the SEPM console and this effectively reduces the workload. Scheme 2: Separate Management In scheme 2, the SEP software is installed on each server but no server has SEPM installed. Each server needs to connect to the Internet to upgrade antivirus engines. You need to install SEP and configure antivirus policies for each server independently. This scheme is of poor manageability. For security and manageability purposes, scheme 1 is recommended. If the system is of a small scale and the customer is much concerned about the cost, scheme 2 is recommended. 57

66 Hardware configurations required by antivirus software: SEP According to related manual of Symantec, it is recommended that the server where SEP is to be installed be configured with 2 GHz CPU, 512 MB memory, and 700 MB disk space. Servers in the system can meet these requirements. SEPM According to related manual of Symantec, it is recommended that the server where SEPM is to be installed be configured with 2 GHz CPU, 2 GB memory, and 4 GB disk space. SEPM can be installed on a server or a desktop computer Network Design DAG Network Each DAG and its members must meet certain network requirements. Each DAG uses an MAPI network, over which other Exchange servers and directory servers communicate with members of this DAG. In addition, a DAG does not use replication networks or use multiple replication networks, which are dedicated to log transfer and seed settings. Members of a DAG can use the same network. It is recommended that each DAG have two networks: an MAPI network and a replication network. This ensures network and network path redundancy and enables the system to distinguish server faults from network faults. If a single NIC is used, the system cannot distinguish server faults from network faults. This document assumes that each DAG member has at least two NICs, each DAG has an MAPI network and a replication network, and the system can distinguish server faults from network faults. Observe the following rules when designing the basic network architecture for the DAG: Each DAG member has at least one NIC that can communicate with all other members. If a single network is used, gigabit Ethernet (GE) is recommended. If each DAG member uses a single NIC, the DAG network must be enabled for replication and be configured as an MAPI network. The system still uses the MAPI network as the replication network because no other network exists. If each DAG member uses a single NIC, it is recommended that a single NIC and a single network path be used in overall system design. If each DAG member uses two NICs, an MAPI network and a replication network can be provided. The details are as follows: If a fault affects the MAPI network, a failover is triggered provided that a healthy mailbox database copy can be activated. If a fault affects the replication network, log transfer and seed setting are switched over to the MAPI network provided that the MAPI network remains unaffected. After the replication network resumes, log transfer and seed setting are automatically switched back to the replication network. Each DAG member must have the same number of networks. For example, if a DAG member uses a single NIC, other DAG members must use a single NIC. A DAG cannot have multiple MAPI networks. The MAPI network must provide connections to other Exchange servers and services such as the AD and DNS services. 58

67 Replication networks can be added as required. The NIC grouping or other similar technologies can be employed to avoid SPOF of NICs. SPOF may still occur on the network even if the NIC grouping technology is employed. Every network of a DAG member must reside on the subnet of the member. The DAG members may reside on different subnets. However, the MAPI and replication networks must route data and provide connections so that: Every network of a DAG member resides on the subnet of the member and this subnet is isolated from the subnets where other networks of the member reside. The MAPI network of a DAG member can communicate with the MAPI networks of other members. The replication network of a DAG member can communicate with the replication networks of other members. No direct route exists between the MAPI network of a DAG member and the replication network of another DAG member. In addition, no direct route exists between replication networks of DAG members. The roundabout delay between DAG members does not exceed 250 ms regardless of the physical locations of the members. If multiple data centers are configured, the roundabout delay requirement may not constitute the strictest network bandwidth and delay requirement. Network requirements must be determined based on the evaluated total network load, including traffic of client access, AD, transmission, and continuous replication. The DAG network supports IPv4 and IPv6. IPv6 is supported only when IPv4 is also used. In other words, the DAG network does not support IPv6 independently. IPv6 addresses are supported only when IPv4 and IPv6 are both enabled and supported. The Exchange Server 2010 enables all servers to receive and send data to devices, servers, and clients that use IPv6 addresses. The automatic private IP addressing (APIPA) is a function provided by Microsoft Windows. The function automatically assigns an IP address when no Dynamic Host Configuration Protocol (DHCP) server is available on the network. The DAG and Exchange Server 2010 do not support APIPA addresses, including APIPA addresses manually assigned. Requirements for DAG Names and IP Addresses When a DAG is created, a unique name is specified. In addition, one or more static IP addresses are assigned to the DAG or the DAG is configured to use DHCP. The static and dynamic IP addresses assigned to DAGs must reside on the MAPI network. A DAG requires at least one IP address on the MAPI network. More IP addresses are required when the MAPI network covers multiple subnets. NIC Configuration for a DAG Each NIC must be correctly configured according to the plan. The configurations of an NIC adapter for the MAPI network are different from those of an NIC for the replication network. The network precedence must be configured in Windows to ensure that the MAPI network if available, is connected first. Configure NICs for the MAPI network based on Table

68 Table 2-28 Configurations of NICs for the MAPI network Item Microsoft Quality of service (QoS) Microsoft file and printer sharing IPv6 (TCP/IPv6) IPv4 (TCP/IPv4) I/O driver of the topology discovery mapping device at the link layer Topology discovery response program at the link layer Setting Network client Packet planning program Enabled Enabled (optional) Enabled Enabled Enabled Configure TCP/IPv4 attributes of NICs for the MAPI network as follows: Manually assign IP addresses or enable DHCP. If DHCP is enabled, set server IP addresses to be permanently used. Use the default gateway though it is not required. Configure at least one DNS server address. It is recommended that multiple DNS servers be configured for redundancy purposes. Select the Register this connection's addresses in DNS check box. Configure NICs for the replication network based on Table Table 2-29 NICs for the replication network Item Microsoft network client QoS packet planning program Microsoft file and printer sharing IPv6 (TCP/IPv6) IPv4 (TCP/IPv4) I/O driver of the topology discovery mapping device at the link layer Topology discovery response program at the link layer Setting Disabled Disabled Disabled Disabled Enabled Enabled Enabled Configure TCP/IPv4 attributes of NICs for the replication network as follows: Manually assign IP addresses or enable DHCP. If DHCP is enabled, set server IP addresses to be permanently used. Typically no default gateway is configured for the replication network. If the MAPI network has a default gateway, other networks should not have. Permanent static routes can be configured on the replication network to route traffic to the networks of other 60

69 DAG members that uses the gateway address. Traffic can be routed among replication networks based on this gateway address. Other traffic that does not match this route is handled by the default gateway on the MAPI network. Do not configure any DNS server address. Do not select the Register this connection's addresses in DNS check box. 61

70 3 Configuration of the Exchange Server Configuration of the Exchange Server HA and Disaster Recovery As mails usually carry important data, the mail system must provide powerful protection functions and HA for law and regulation compliance and internal audit. This is to ensure that all important information is transmitted smoothly. The continuous replication mechanism is introduced in the Exchange Server 2010 so that mails are backed up and stored to remote databases. This eliminates the need for additional storage systems and reduces the cost in deploying a complete mail system. In addition, the mailbox resiliency mechanism is introduced for disaster recovery. A maximum of 16 database copies are available for writing backup data. This ensures that mailboxes can be recovered from disasters by the automated restoration program. Function description: DAG A DAG is a group of MBSs and provides disk-level, server-level, and data center-level restoration capabilities based on continuous replication. Database-level failover A DAG in the Exchange Server 2010 provides database-level failover, which is easier than failover in traditional clusters. Database-level interruption, such as a disk fault, does not affect users. Databases are not bound with servers and therefore the Huawei Exchange system can switch between database copies easily when a disk fault occurs. In addition, the failover time is shortened to 30 seconds and this effectively improves the working efficiency of the organization. Improved site resiliency The DAG extends data replication among data centers to that between archive sites. This improves site resiliency. Log files are compressed to shorten the transmission time and reduce the network bandwidth usage. Easy deployment After the Exchange system is deployed initially, the administrator can configure a DAG for HA purposes, without a need for server reinstallation. For small-scale organizations, two servers are enough for deploying the CAS and HTS and implementing full redundancy. HA can therefore be easily implemented. Cluster management The DAG has the failover function, which is easier than failover in traditional clusters. Functions of Windows clusters are integrated with the Exchange Server The mail 62

71 3 Configuration of the Exchange Server 2010 system of an enterprise can run properly even if the administrator does not know about clusters or usage of independent management tools. Backup-free recovery The DAG structure allows delay in replaying log files so that the administrator can recover a database to a certain time point without tapes. The organization can recover from a failure depending on its HA infrastructure, instead of data backup to tapes. This significantly reduces the operating expense (OPEX). Transmission recovery The native protection function of transport servers helps avoid message queue loss caused by disk or server failures. After a mail is forwarded to the next hop in the organization, the transport server saves a shadow copy of this mail. If a fault occurs in subsequent hops before delivery success is reported, the mail is resubmitted along a different path. Online mailbox move The administrator can move a mailbox between databases without affecting the user. During an online mailbox move, the user can still access the mailbox to send and receive mails. This enables the administrator to perform system maintenance flexibly during working hours, instead of at night or weekends. 3.2 Restrictions on Mailboxes MBSs in the Exchange Server 2010 host mailbox databases, store mails, and provide advanced planning services for Microsoft Office Outlook users. Restrictions on mailbox and mail sizes help the administrator perform Exchange database management, backup, and restoration. The following is an example of setting restrictions on the mailbox size: 1. Create storage groups and databases for all users. Limit the mailbox size to 500 MB. A warning mail is sent when a mailbox reaches 460 MB. Mail sending is prohibited when a mailbox reaches 480 MB. Mail sending and receiving are prohibited when a mailbox reaches 500 MB. Figure 3-1 Mailbox restrictions for all users 63

72 3 Configuration of the Exchange Server Create a storage group and a database for VIP users. Limit the mailbox size to 1000 MB. A warning mail is sent when a mailbox reaches 800 MB. Mail sending is prohibited when a mailbox reaches 900 MB. Mail sending and receiving are prohibited when a mailbox reaches 1000 MB. Figure 3-2 Mailbox restrictions for VIP users The different types of restrictions are listed as follows: Organization restrictions: The restrictions apply to all Exchange servers in an organization. Specified mail restrictions apply to all servers that have the HTS installed. Connector restrictions: The restrictions apply to mails that are transmitted by specified sending connectors, receiving connectors, or external connectors. Connectors can be defined on the HTS or ETS. Server restrictions: The restrictions apply to specified HTSs or ETSs. Specified mail restrictions are not stored in the AD service. Mail restrictions can be independently defined on each HTS or ETS. User restrictions: The restrictions apply to certain objects such as mailboxes, contacts, contact groups, or public folders. According to research, the recommended maximum size of received and sent mails is 10 MB. You can run the Set-ReceiveConnector and Set-SendConnector commands to set the maximum size of received and sent mails to 10 MB for receiving and sending connectors. 3.3 Management Mode The Exchange Server 2010 has different types of administrators defined for refined management. The details are as follows: Organization administrator: This administrator has full control rights for all attributes and objects in an Exchange organization. Public folder administrator: This administrator has permission to manage public folders. 64

73 3 Configuration of the Exchange Server 2010 Recipient administrator: This administrator has permission to modify AD users, contacts, contact groups, dynamic contact groups, and public folders. Browse administrator: This administrator has read-only permission for the Exchange organization tree under the AD and all other Microsoft Windows domain containers containing Exchange recipients. Server administrator: This administrator has permission to access local Exchange server configurations in the AD or those on physical servers that have the AD installed. According to research, the system administrator of the Exchange Server 2010 has the operation rights of the server administrator, recipient administrator, and public folder administrator. 3.4 Client Access Mode The increasing mobility of users raises an urgent need for remote mailbox access The Exchange Server 2010 provides multiple client-specific functions to improve user experience and enable the administrator to easily manage the mail system. It provides the latest OWA to better service experience of end users and improve working efficiency. The Outlook Web App 2010 delivers a pack of new functions and function enhancements. New functions, including intelligent meeting arrangement, Windows SharePoint Services, Windows file sharing, and mobile device management, are now available for users. The OWA presents enhancements of mail functions such as the search, reminder, and address book. The Exchange Server 2003 and later versions allow users to access information on the Internet through the Windows RPC over HTTP component. Windows RPC over HTTP involves RPC at the HTTP layer. The RPC port does not need to be enabled and the traffic can still traverse the network firewall. The Exchange Server 2010 builds on this function and simplifies its deployment and management. It is recommended that the OWA be enabled for all users and Outlook Anywhere be enabled for laptop users. By default, the OWA is enabled after the CAS is installed. For laptop users, RPC over HTTP is installed on the CAS and Outlook Anywhere is enabled on the management console of the Exchange Server System Security Account and Group Security Secure Access Independent accounts and groups need to be created and added to the local administrator group and organization administrators group. By default, mails are transmitted in cipher text within an Exchange system. Remote users can establish security connections with MBSs in the following ways: Establishing virtual private network (VPN) connections with MBSs. Establishing SSL connections with MBSs through RPC over HTTP. Establishing SSL connections with MBSs through POP3, SMTP, or IMAP ports. 65

74 3 Configuration of the Exchange Server 2010 Mail Security Mail Approval Establishing SSL connections with MBSs over HTTP using a browser. Establishing SSL connections with MBSs by using the ActiveSync on an intelligent device such as a PocketPC Phone or Smartphone. All secure mail clients support robust authentication and encryption between MBSs and mail clients. This protects mails between MBSs and mail clients from hackers. Identity authentication is performed based on digital signatures. This also prevents users from denying association with signed contents. The Exchange Server 2010 supports the Secure/Multipurpose Internet Mail Extensions (S/MIME) protocol, which is based on industrial standards. Users can encrypt mails and add digital signatures to mails on S/MIME-enabled clients such as Outlook, Outlook Web Access, and Outlook Express clients. The Exchange Server 2010 integrates the AD service to strictly control user access rights. This is to prevent illegitimate access to others' mailboxes and forged mails, thereby ensuring system security and eliminating security risks caused by defects of protocols. Outlook clients provide encryption functions to protect local personal folders. Therefore, it is recommended that: A password be set for security. Mails be stored to a local computer for convenience. Folders be cleaned up and compressed, and unwanted mails and outdated mails with large attachments be deleted. Mail approval is a new function of the Exchange Server With this function, mails bound for certain users can be forwarded only after being approved. This function is used to control skip-level reporting. Mail approval is implemented based on server rules. A mail is forwarded to the approver's mailbox before being forwarded to the destination. The approver can view the whole mail text and approve, reject, or comment a mail. Only approved mails are forwarded to their destination. Rejected mails with comments are returned to the senders. 66

75 4 Configuration Lists 4 Configuration Lists 4.1 Configuration List for the Low-cost 600-user Solution Hardware Configurations Table 4-1 Hardware configurations for the low-cost 600-user solution Server Hardware Configuration Quantity Remarks Multi-role server (MBS, HTS, and CAS) RH2285 V2 configured with a quad-core Xeon E CPU, three 8 GB dual in-line memory modules (DIMMs), eight 7.2K RPM 1 TB SATA disks, and a dual-port GE NIC 2 Two disks are configured as a RAID 1 group for deploying the operating system, and six function as data storage and maintenance disks. LSISAS2208 RAID controller cards are used. ETS Directory server RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks 2 Two disks are configured as a RAID 1 group for deploying the operating system. LSISAS2308 RAID controller cards are used. 2 Two disks are configured as a RAID 1 group for deploying the operating system. LSISAS2308 RAID controller cards are used. Load balancing F5 BIG-IP LTM Two F for CAS,other two F for TMG server 67

76 4 Configuration Lists Server Hardware Configuration Quantity Remarks Backup server Archiving server Tape library TMG server Manager server RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, twelve 7.2K RPM 1 TB SATA disks, and a dual-port NIC RH2285 V2 configured with two quad-core Xeon E CPUs, two 8 GB DIMMs, twelve 7.2K RPM 2 TB SATA disks, at least two GE ports, and an optional dual-port 8 Gbit/s FC HBA card A Quantum Scalar i40 (LTO-5 tapes, 15 m multimode fibers, and two tape drives) RH2285 V2 configured with a quad-core Xeon E CPU, two 8 GB DIMMs, two 10K RPM 300 GB SAS disks, and a dual-port NIC RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks 1 Two disks are configured as a RAID 1 group for deploying the operating system, eight are configured as RAID 5 groups, one functions as a hot spare disk, and one functions as a maintenance disk. LSISAS2208 RAID controller cards are used. 1 Two disks are configured as a RAID 1 group, eight are configured as RAID 5 groups, one functions as a hot spare disk, and one functions as a maintenance disk. LSISAS2208 RAID controller cards are used. If a tape library is configured, an 8 Gbit/s FC card is required. 1 Height: 3 U (5.25 in. or mm) 2 Two disks are configured as a RAID 1 group for deploying the operating system, TMG system, and Web cache. LSISAS2308 RAID controller cards are used. 1 The manager server can be deployed on a desktop computer. Switch 48 GE switch 2 FC: fiber channel HBA: host bus adapter 68

77 4 Configuration Lists Software Configurations Table 4-2 Software configurations for the low-cost 600-user solution Item Software Configuration Function Quantity Multi-role server (MBS, HTS, and CAS) Windows Server 2008 R2 Enterprise Edition Microsoft Exchange Server 2010 Enterprise Edition Hosting mailboxes. Each MBS hosts a maximum of 600 mailboxes with HA configuration. 2 ETS Windows Server 2008 R2 Standard Edition Internet access control, antivirus, and antispam. 2 Microsoft Exchange Server 2010 Enterprise Edition Directory server Windows Server 2008 R2 Standard Edition Global account and resource management. The directory server also provides functions required by the MBS. The information is automatically synchronized among organizations. 2 Standard client access license Exchange Server CAL Standard client access licenses 600 Enterprise client access license Exchange Server CAL Enterprise client access licenses 600 Antispam and antivirus Forefront Protection 2010 for Exchange Antivirus and antispam. The software can be provided free of charge together with enterprise client access licenses. 4 Backup server Client agent Windows Server 2008 R2 Enterprise Edition SYMC NETBACKUP SERVER 7.5 WIN/LNX/SOLX64 1 SERVER TIER 1 STD LIC SYMC NETBACKUP CLIENT APPLICATION AND DATABASE PACK 7.5 WIN/LNX/SOLX64 1 SERVER TIER 1 STD LIC Backup server software. 1 Backup client agent. 2 Client license SYMC NETBACKUP STANDARD CLIENT 7.5 XPLAT 1 SERVER STD LIC Client licenses. 2 69

78 4 Configuration Lists Item Software Configuration Function Quantity Archiving server Query software (optional) Windows Server 2008 R2 Standard Edition SYMC ENTERPRISE VAULT STORAGE MANAGEMENT FOR MICROSOFT EXCHANGE 10.0 WIN PER USER STD LIC Microsoft SQL Server 2008 R2, English, Enterprise Edition, 10 CAL, for Windows, 64-bit, SQL2008R2 Ent OEM COA Lic, SQL2008R2 CD, No Doc, Without Product Services SYMC ENTERPRISE VAULT E-DISCOVERY STANDARD EDITION FOR MICROSOFT EXCHANGE 10.0 WIN PER USER STD LIC Archiving server license. 600 Querying archived mails. 600 TMG server Windows Server 2008 R2 Enterprise Edition x64 Microsoft Forefront Threat Management Gateway 2010 Enterprise Edition Network Inspection System (NIS) software license and Web Protection software license, which are optional. 2 System protection Choice 1: SEP Server Edition 7 x 24 services for 12 months (SYMC ENDPOINT PROTECTION FOR NETWORK SERVERS 12.1 PER SERVER BNDL STD LIC EXPRESS BAND D ESSENTIAL 12 MONTHS) Protecting servers against viruses. 14 (equal to the server quantity) Choice 2: SEP Server Edition 7 x 24 services for 36 months (SYMC ENDPOINT PROTECTION FOR NETWORK SERVERS 12.1 PER SERVER BNDL STD LIC EXPRESS BAND D ESSENTIAL 36 MONTHS) Operating system of the manager server: Windows Server 2008 R2 Standard Edition Link Configurations The bandwidth of the replication network must be ensured so that database copies are synchronized quickly. 70

79 4 Configuration Lists Size of a daily transaction log file = Number of mails received/sent per day x Average mail size x Number of mailboxes = 200 (mails) x 75 KB x 600 (mailboxes) = 8.5 GB If there are eight working hours per day, the bandwidth required by the replication network is calculated as follows: 8.5 x 1024 x 8/(8 x 3600) = 2.4 Mbit/s It is recommended that at least 4 Mbit/s of bandwidth be ensured for the service network because the service network is responsible for log replication when the replication network is unavailable. You are advised to configure one more NIC for each MBS. 4.2 Configuration List for the Low-cost 1200-user Solution Hardware Configurations Table 4-3 Hardware configurations for the low-cost 1200-user solution Server Hardware Configuration Quantity Remarks Multi-role server (MBS, HTS, and CAS) ETS Directory server RH2285 V2 configured with two quad-core Xeon E CPUs, four 8 GB DIMMs, two 10K RPM 300 GB SAS disks, ten 7.2K RPM 2 TB SATA disks, and a dual-port GE NIC RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks 2 The two SAS disks are configured as a RAID 1 group for deploying the operating system, and the ten SATA disks function as data storage and maintenance disks. LSISAS2208 RAID controller cards are used. 2 Two disks are configured as a RAID 1 group for deploying the operating system. LSISAS2308 RAID controller cards are used. 2 Two disks are configured as a RAID 1 group for deploying the operating system. LSISAS2308 RAID controller cards are used. Load balancing F5 BIG-IP LTM Two F for CAS,other two F for TMG server 71

80 4 Configuration Lists Server Hardware Configuration Quantity Remarks Backup server Archiving server Database server Tape library TMG server Manager server RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, twelve 7.2K RPM 2 TB SATA disks, and two GE ports RH2285 V2 configured with two quad-core Xeon E CPUs, two 8 GB DIMMs, twelve 7.2K RPM 2 TB SATA disks, at least two GE ports, and a dual-port 8 Gbit/s FC HBA card RH2285 V2 configured with two quad-core Xeon E CPUs, two 8 GB DIMMs, four 7.2K RPM 1 TB SATA disks, and at least two GE ports A Quantum Scalar i40 (two LTO-5 tapes, 41 active slots, two 15 m multimode fibers, and two tape drives) RH2285 V2 configured with two quad-core Xeon E CPUs, two 8 GB DIMMs, two 10K RPM 300 GB SAS disks, and a dual-port GE NIC RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks 1 Two disks are configured as a RAID 1 group for deploying the operating system, eight are configured as RAID 5 groups, one functions as a hot spare disk, and one functions as a maintenance disk. LSISAS2208 RAID controller cards are used. 1 Two disks are configured as a RAID 1 group, eight are configured as RAID 5 groups, one functions as a hot spare disk, and one functions as a maintenance disk. LSISAS2208 RAID controller cards are used.if a tape library is configured, an 8 Gbit/s FC card is required. 1 The four disks are configured as two RAID 1 groups respectively for deploying the operating system and SQL Server. LSISAS2208 RAID controller cards are used. 1 Height: 5 U (8.75 in. or mm) 2 The two disks are configured as a RAID 1 group for deploying the operating system, TMG system, and Web cache. LSISAS2308 RAID controller cards are used. 1 The manager server can be deployed on a desktop computer. Switch 48 GE switch 2 72

81 4 Configuration Lists Software Configurations Table 4-4 Software configurations for the low-cost 1200-user solution Item Software Configuration Function Quantity Multi-role server (MBS, HTS, and CAS) Windows Server 2008 R2 Enterprise Edition Microsoft Exchange Server 2010 Enterprise Edition Hosting mailboxes. Each MBS hosts a maximum of 1200 mailboxes with HA configuration. Transmission control and client access. 2 ETS Windows Server 2008 R2 Standard Edition Internet access control, antivirus, and antispam. 2 Microsoft Exchange Server 2010 Enterprise Edition Directory server Windows Server 2008 R2 Standard Edition Global account and resource management. The directory server also provides functions required by the MBS. The information is automatically synchronized among organizations. 2 Standard client access license Exchange Server CAL Standard client access licenses 1200 Enterprise client access license Exchange Server CAL Enterprise client access licenses 1200 Backup server Client agent Client license Windows Server 2008 R2 Enterprise Edition SYMC NETBACKUP SERVER 7.5 WIN/LNX/SOLX64 1 SERVER TIER 1 STD LIC SYMC NETBACKUP CLIENT APPLICATION AND DATABASE PACK 7.5 WIN/LNX/SOLX64 1 SERVER TIER 1 STD LIC SYMC NETBACKUP STANDARD CLIENT 7.5 XPLAT 1 SERVER STD LIC Backup server software. 1 Backup client agent. 2 Client licenses. 2 73

82 4 Configuration Lists Item Software Configuration Function Quantity Archiving server Windows Server 2008 R2 Standard Edition SYMC ENTERPRISE VAULT STORAGE MANAGEMENT FOR MICROSOFT EXCHANGE 10.0 WIN PER USER STD LIC Archiving server license Query software (optional) SYMC ENTERPRISE VAULT E-DISCOVERY STANDARD EDITION FOR MICROSOFT EXCHANGE 10.0 WIN PER USER STD LIC Querying archived mails Database server Windows Server 2008 R2 Standard Edition Microsoft SQL Server 2008 R2, English, Enterprise Edition, 10 CAL, for Windows, 64-bit, SQL2008R2 Ent OEM COA Lic, SQL2008R2 CD, No Doc, Without Product Services Archiving database. 1 TMG server Windows Server 2008 R2 Enterprise Edition x64 Microsoft Forefront Threat Management Gateway 2010 Enterprise Edition NIS software license and Web Protection software license, which are optional. 2 System protection Choice 1: SEP Server Edition 7 x 24 services for 12 months (SYMC ENDPOINT PROTECTION FOR NETWORK SERVERS 12.1 PER SERVER BNDL STD LIC EXPRESS BAND D ESSENTIAL 12 MONTHS) Choice 2: SEP Server Edition 7 x 24 services for 36 months (SYMC ENDPOINT PROTECTION FOR NETWORK SERVERS 12.1 PER SERVER BNDL STD LIC EXPRESS BAND D ESSENTIAL 36 MONTHS) Operating system of the manager server: Windows Server 2008 R2 Standard Edition For defense against viruses, thereby protecting server security. 12 (equal to the server quantity) 1 74

83 4 Configuration Lists Link Configurations The bandwidth of the replication network must be ensured to ensure quick data synchronization between database copies. Size of a daily transaction log file = Number of mails received/sent per day x Average mail size x Number of mailboxes = 200 x 75 KB x 1200 = 17 GB If there are eight working hours per day, the bandwidth required by the replication network is calculated as follows: 17 x 1024 x 8/(8 x 3600) = 4.84 Mbit/s It is recommended that at least 8 Mbit/s of bandwidth be ensured for the service network because the service network is responsible for log replication when the replication network is unavailable. You are advised to configure one more NIC for each MBS. 4.3 Configuration List for the High-performance 1200-user Solution Hardware Configurations Table 4-5 Hardware configurations for the high-performance 1200-user solution Server Hardware Configuration Quantity Remarks Multi-role server (MBS, HTS, and CAS) RH2285 V2 configured with two quad-core Xeon E CPUs, eight 8 GB DIMMs, four 7.2K RPM 1 TB SATA disks, a dual-port GE NIC, and two 640 GB SSD card 2 Two disks are configured as a RAID 1 group for deploying the operating system, one functions as a hot spare disk, and one functions as a maintenance disk. LSISAS2308 RAID controller cards are used. ETS Directory server RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks 2 Two disks are configured as a RAID 1 group for deploying the operating system. LSISAS2308 RAID controller cards are used. 2 Two disks are configured as a RAID 1 group for deploying the operating system. LSISAS2308 RAID controller cards are used. Load balancing F5 BIG-IP LTM Two F for CAS,other two F for TMG server 75

84 4 Configuration Lists Server Hardware Configuration Quantity Remarks Backup server Archiving server Database server RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, twelve 7.2K RPM 2 TB SATA disks, and two GE ports RH2285 V2 configured with two quad-core Xeon E CPUs, two 8 GB DIMMs, twelve 7.2K RPM 2 TB SATA disks, at least two GE ports, and a dual-port 8 Gbit/s FC HBA card RH2285 V2 configured with two quad-core Xeon E CPUs, two 8 GB DIMMs, four 7.2K RPM 1 TB SATA disks, and at least two GE ports 1 Two disks are configured as a RAID 1 group for deploying the operating system, eight are configured as RAID 5 groups, one functions as a hot spare disk, and one functions as a maintenance disk. LSISAS2208 RAID controller cards are used. 1 Two disks are configured as a RAID 1 group, eight are configured as RAID 5 groups, one functions as a hot spare disk, and one functions as a maintenance disk. LSISAS2208 RAID controller cards are used. If a tape library is configured, an 8 Gbit/s FC card is required. 1 The four disks are configured as two RAID 1 groups respectively for deploying the operating system and SQL Server. LSISAS2208 RAID controller cards are used. If a tape library is configured, an 8 Gbit/s FC card is required. Tape library A Quantum Scalar i40 (two LTO-5 tapes, 41 active slots, two 15 m multimode fibers, and two tape drives) TMG server RH2285 V2 configured with two quad-core Xeon E CPUs, two 8 GB DIMMs, two 10K RPM 300 GB SAS disks, and at least two GE ports 1 Height: 5 U (8.75 in. or mm) 2 The two disks are configured as a RAID 1 group for deploying the operating system, TMG system, and Web cache. LSISAS2308 RAID controller cards are used. Manager server RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks 1 The manager server can be deployed on a desktop computer. Switch 48 GE switch 2 76

85 4 Configuration Lists Software Configurations Table 4-6 Software configurations for the high-performance 1200-user solution Item Software Configuration Function Quantity Multi-role server (MBS, HTS, and CAS) Windows Server 2008 R2 Enterprise Edition Microsoft Exchange Server 2010 Enterprise Edition Hosting mailboxes. Each MBS hosts a maximum of 1200 mailboxes with HA configuration. 2 Transmission control and client access. ETS Windows Server 2008 R2 Standard Edition Microsoft Exchange Server 2010 Enterprise Edition Internet access control, antivirus, and antispam. 2 Directory server Windows Server 2008 R2 Standard Edition Global account and resource management. The directory server also provides functions required by the MBS. The information is automatically synchronized among organizations. 2 Standard client access license Exchange Server CAL Standard client access licenses 1200 Enterprise client access license Exchange Server CAL Enterprise client access licenses 1200 Backup server Windows Server 2008 R2 Enterprise Edition Backup server software. 1 SYMC NETBACKUP SERVER 7.5 WIN/LNX/SOLX64 1 SERVER TIER 1 STD LIC Client agent Client license SYMC NETBACKUP CLIENT APPLICATION AND DATABASE PACK 7.5 WIN/LNX/SOLX64 1 SERVER TIER 1 STD LIC SYMC NETBACKUP STANDARD CLIENT 7.5 XPLAT 1 SERVER STD LIC Backup client agent. 2 Client licenses. 2 77

86 4 Configuration Lists Item Software Configuration Function Quantity Archiving server Windows Server 2008 R2 Standard Edition Archiving server license SYMC ENTERPRISE VAULT STORAGE MANAGEMENT FOR MICROSOFT EXCHANGE 10.0 WIN PER USER STD LIC Query software (optional) SYMC ENTERPRISE VAULT E-DISCOVERY STANDARD EDITION FOR MICROSOFT EXCHANGE 10.0 WIN PER USER STD LIC Querying archived mails Database server Windows Server 2008 R2 Standard Edition Microsoft SQL Server 2008 R2, English, Enterprise Edition, 10 CAL, for Windows, 64-bit, SQL2008R2 Ent OEM COA Lic, SQL2008R2 CD, No Doc, Without Product Services Archiving database. 1 TMG server Windows Server 2008 R2 Enterprise Edition x64 Microsoft Forefront Threat Management Gateway 2010 Enterprise Edition NIS software license and Web Protection software license, which are optional. 2 System protection Choice 1: SEP Server Edition 7 x 24 services for 12 months (SYMC ENDPOINT PROTECTION FOR NETWORK SERVERS 12.1 PER SERVER BNDL STD LIC EXPRESS BAND D ESSENTIAL 12 MONTHS) Protecting servers against viruses. 12 (equal to the server quantity) Choice 2: SEP Server Edition 7 x 24 services for 36 months (SYMC ENDPOINT PROTECTION FOR NETWORK SERVERS 12.1 PER SERVER BNDL STD LIC EXPRESS BAND D ESSENTIAL 36 MONTHS) Operating system of the manager server: Windows Server 2008 R2 Standard Edition Link Configurations The bandwidth of the replication network must be ensured so that database copies are synchronized quickly. 78

87 4 Configuration Lists Size of a daily transaction log file = Number of mails received/sent per day x Average mail size x Number of mailboxes = 500 (mails) x 75 KB x 1200 (mailboxes) = 43 GB If there are eight working hours per day, the bandwidth required by the replication network is calculated as follows: 43 x 1024 x 8/(8 x 3600) = Mbit/s It is recommended that at least 16 Mbit/s of bandwidth be ensured for the service network because the service network is responsible for log replication when the replication network is unavailable. You are advised to configure one more NIC for each MBS. 4.4 Configuration List for the Low-cost 3000-user Solution Hardware Configurations Table 4-7 Hardware configurations for the low-cost 3000-user solution Server Hardware Configuration Quantity Remarks Multi-role server (MBS, HTS, and CAS) ETS Directory server RH2285 V2 configured with two six-core Xeon E CPUs, six 8 GB DIMMs, two 10K RPM 300 GB SAS disks, twelve 7.2K RPM 2 TB SATA disks, and a dual-port GE NIC RH2285 V2 configured with a six-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks RH2285 V2 configured with a six-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks 3 The two SAS disks are configured as a RAID 1 group for deploying the operating system, and the twelve SATA disks function as data storage and maintenance disks. LSISAS2208 RAID controller cards are used. 2 Two disks are configured as a RAID 1 group for deploying the operating system. LSISAS2308 RAID controller cards are used. 2 Two disks are configured as a RAID 1 group for deploying the operating system. LSISAS2308 RAID controller cards are used. Load balancing F5 BIG-IP LTM 1600/ Two F for CAS,other two F for TMG server 79

88 4 Configuration Lists Server Hardware Configuration Quantity Remarks Backup server RH2285 V2 server configured with a quad-core Xeon E CPU, an 8 GB DIMM, twelve 7.2K RPM 3 TB SATA disks, and an NIC with two network ports (NIC bonding is supported in Windows) 1 Two disks are configured as a RAID 1 group for deploying the operating system, eight are configured as RAID 5 groups, one functions as a hot spare disk, and one functions as a maintenance disk. LSI2208 RAID controller card Archiving server Database server Tape library TMG server Manager server RH2285 V2 configured with two quad-core Xeon E CPUs, two 8 GB DIMMs, twelve 7.2K RPM 2 TB SATA disks, at least two GE ports, and a dual-port 8 Gbit/s FC HBA card RH2285 V2 configured with two quad-core Xeon E CPUs, two 8 GB DIMMs, eight 7.2K RPM 1 TB SATA disks, and at least two GE ports A Quantum Scalar i40 (two LTO-4 tapes, 87 active slots, two 15 m multimode fibers, and two tape drives) RH2285 V2 configured with two quad-core Xeon E CPUs, four 8 GB DIMMs, three 10K RPM 300 GB SAS disks, and at least two GE ports RH2285 V2 configured with a quad-core Xeon E CPU, an 8 GB DIMM, and two 10K RPM 300 GB SAS disks 1 Two disks are configured as a RAID 1 group, eight are configured as RAID 5 groups, one functions as a hot spare disk, and one functions as a maintenance disk. LSISAS2208 RAID controller cards are used.if a tape library is configured, an 8 Gbit/s FC card is required. 1 Two disks are configured as a RAID 1 group for deploying the operating system, five are configured as a RAID 5 group for deploying the SQL Server, and one functions as a hot spare disk. LSISAS2208 RAID controller cards are used. 1 Height (including extension enclosures): 14 U (24.5 in. or 623 mm) 2 The three disks are configured as a RAID 5 group for deploying the operating system, TMG system, and Web cache. LSISAS2208 RAID controller cards are used. 1 The manager server can be deployed on a desktop computer. Switch 48 GE switch 2 80

89 4 Configuration Lists Software Configurations Table 4-8 Software configurations for the low-cost 3000-user solution Item Software Configuration Function Quantity Multi-role server (MBS, HTS, and CAS) Windows Server 2008 R2 Enterprise Edition Microsoft Exchange Server 2010 Enterprise Edition Hosting mailboxes. Each MBS hosts a maximum of 3000 mailboxes with HA configuration. Transmission control and client access. 3 ETS Windows Server 2008 R2 Standard Edition Internet access control, antivirus, and antispam. 2 Microsoft Exchange Server 2010 Enterprise Edition Directory server Windows Server 2008 R2 Standard Edition Global account and resource management. The directory server also provides functions required by the MBS. The information is automatically synchronized among organizations. 2 Standard client access license Exchange Server CAL Standard client access licenses 3000 Enterprise client access license Exchange Server CAL Enterprise client access licenses 3000 Backup server Windows Server 2008 R2 Enterprise Edition SYMC NETBACKUP SERVER 7.5 WIN/LNX/SOLX64 1 SERVER TIER 1 STD LIC Backup server software. 1 Client agent Client license SYMC NETBACKUP CLIENT APPLICATION AND DATABASE PACK 7.5 WIN/LNX/SOLX64 1 SERVER TIER 1 STD LIC SYMC NETBACKUP STANDARD CLIENT 7.5 XPLAT 1 SERVER STD LIC Backup client agent. 3 Client licenses. 3 81

90 4 Configuration Lists Item Software Configuration Function Quantity Archiving server Query software (optional) Database server Windows Server 2008 R2 Standard Edition SYMC ENTERPRISE VAULT STORAGE MANAGEMENT FOR MICROSOFT EXCHANGE 10.0 WIN PER USER STD LIC SYMC ENTERPRISE VAULT E-DISCOVERY STANDARD EDITION FOR MICROSOFT EXCHANGE 10.0 WIN PER USER STD LIC Windows Server 2008 R2 Standard Edition Microsoft SQL Server 2008 R2, English, Enterprise Edition, 10 CAL, for Windows, 64-bit, SQL2008R2 Ent OEM COA Lic, SQL2008R2 CD, No Doc, Without Product Services Archiving server license Querying archived mails Archiving database. 1 TMG server Windows Server 2008 R2 Enterprise Edition x64 Microsoft Forefront Threat Management Gateway 2010 Enterprise Edition NIS software license and Web Protection software license, which are optional. 2 System protection Choice 1: SEP Server Edition 7 x 24 services for 12 months (SYMC ENDPOINT PROTECTION FOR NETWORK SERVERS 12.1 PER SERVER BNDL STD LIC EXPRESS BAND D ESSENTIAL 12 MONTHS) Protecting servers against viruses. 13 (equal to the server quantity) Choice 2: SEP Server Edition 7 x 24 services for 36 months (SYMC ENDPOINT PROTECTION FOR NETWORK SERVERS 12.1 PER SERVER BNDL STD LIC EXPRESS BAND D ESSENTIAL 36 MONTHS) Operating system of the manager server: Windows Server 2008 R2 Standard Edition 1 82

91 4 Configuration Lists Link Configurations The bandwidth of the replication network must be ensured so that database copies are synchronized quickly. Size of a daily transaction log file = Number of mails received/sent per day x Average mail size x Number of mailboxes = 200 (mails) x 75 KB x 3000 (mailboxes) = 43 GB If there are eight working hours per day, the bandwidth required by the replication network is calculated as follows: 43 x 1024 x 8/(8 x 3600) = Mbit/s It is recommended that at least 16 Mbit/s of bandwidth be ensured for the service network because the service network is responsible for log replication when the replication network is unavailable. You are advised to configure one more NIC for each MBS. 83

92 5 Involved Products 5 Involved Products 5.1 Huawei RH2285 V2 The RH2285 V2 is 2 U high and can be installed in a standard 19-inch cabinet. It comprises the main board, RAID controller card, I/O daughter card, power supply units (PSUs), fans, disks, modules, and backplane. The front panel of the RH2285 V2 provides system buttons and indicators for users to conveniently maintain hard disks and the DVD-ROM drive. Figure 5-1 shows the front view of the RH2285 V2-8S. Figure 5-1 Front view of the RH2288 V2-8S 1 User identity (UID) button/indicator 2 Healthy indicator 3 Ethernet port indicator 1 4 Ethernet port indicator 2 84

93 5 Involved Products 5 Power button/indicator 6 VGA port 7 Embedded DVD-ROM (optional if eight disks are configured) 9 Disk 10 USB port 8 Label (for querying hardware configurations of the RH2285 V2) The rear panel of the RH2285 V2 provides external interfaces and system components. Figure 5-2 shows the rear view of the RH2285 V2. Figure 5-2 Rear view of the RH2285 V2 1 Data transmission indicator 2 Connection indicator 3 Ground screw 4 DB-9 serial port 5 DB-15 VGA port 6 UID indicator 7 RJ45 management port for the baseboard management controller (BMC) 8 GE port 1 9 GE port 2 10 USB port 11 Power socket for PSU 2 12 Indicator for PSU 2 13 PSU 2 14 Power socket for PSU 1 15 Indicator for PSU 1 16 PSU inch disk (optional if 12 disks are configured) inch disk (optional if 12 disks are configured) Table 5-1 lists the specifications of the RH2285 V2. Table 5-1 Specifications of the RH2285 V2 Item Subrack (H x W x D) Specifications 87.5 mm x 448 mm x 700 mm (3.44 in. x in. x in.) 85