WHITE PAPER: ENTERPRISE SOLUTIONS

Transcription

1 WHITE PAPER: ENTERPRISE SOLUTIONS Veritas Storage Foundation HA for Windows by Symantec Maintaining Optimum Availability in Microsoft Exchange Environments

2 White Paper: Symantec Enterprise Solutions Veritas Storage Foundation HA for Windows by Symantec Maintaining Optimum Availability in Microsoft Exchange Environments Contents Executive summary Overview Service-level expectations are extremely high There are many potential sources of problems The application environment is in constant change Backup and recovery isn t enough The need for a unified, comprehensive approach to Exchange availability The Symantec solution for ensuring Exchange availability Solution components Creating resilient storage for Exchange Creating highly available storage configurations Managing growth, nondisruptively Recovering quickly from logical errors Clustering Exchange servers for protection from component failures Cluster configurations Service groups Protecting Exchange from localized failures with campus clusters Protecting Exchange with replicated data clusters Disaster recovery with global clustering Primary and secondary sites Veritas Volume Replicator Option Scenario: Site migration Conclusion

3 Executive summary plays an increasingly important role in most organizations. As an IT administration responsibility, maintaining user access to has attained business-wide importance and visibility. Maintaining Exchange availability requires the close cooperation of the Microsoft Exchange administrator with the IT staff responsible for other application infrastructure components. The range of potential risks to an environment is large, including viruses, component failures, and site-wide disasters. Rather than implementing a variety of point solutions and strategies for each potential risk, offers a single, integrated solution for Exchange availability. Organizations can implement options and agents that meet their specific demands, while building a highly available application and storage infrastructure to support future needs. This paper describes the specific challenges of maintaining availability for Microsoft Exchange, both Exchange 2000 and 2003 environments, and then describes how Veritas Storage Foundation HA for Windows addresses these risks. Overview Microsoft Exchange administration is one of IT s most demanding and thankless tasks. The volume of grows daily, and its importance to business operations is unquestioned. At the same time, most users take access and performance for granted, as a necessary prerequisite for doing business. Exchange administrators receive little attention for maintaining superior service around the clock, but if there s a failure of any type, everyone notices. The problem is broader than IT administration protecting vital services should be a business priority in much the same way as protecting valuable manufacturing or order fulfillment systems. has become an essential business process, and for many companies, the cost of significant downtime is unacceptable. Maintaining highly available service with Microsoft Exchange is a challenging proposition for several reasons. 4

4 Service-level expectations are extremely high Many IT organizations maintain service-level agreements (SLAs) with their users, quantifying acceptable downtime (if any) and response times to any problems that do occur. Meeting these SLAs is important for both the IT organization and the business units represented in the agreements. SLAs and expectations often influence Exchange configuration decisions. For example, many businesses segregate users into separate storage groups and databases within an Exchange server to speed recovery time if a failure can be isolated. Some create Executive storage groups so that the executive-level staff might have service restored more quickly than the rest of the business. There are many potential sources of problems There are many potential sources of failure that could compromise the Exchange server itself, its data, or user access to . In addition to the obvious problems that Exchange administrators worry about (viruses and worms), access is subject to a number of other risks, including: Component failures: A storage subsystem, network router, or server s power supply could fail. Logical data problems: Viruses, user errors, or application problems can result in data corruption or loss. Localized disasters: A power failure or act of terrorism can affect a local data center. Major disasters: Severe weather or earthquakes can disrupt entire geographic regions for a prolonged period of time. Exchange administrators don t have expertise in all of the components that could potentially fail, so they rely on other IT groups to respond to (and prevent) component problems. Disaster recovery planning clearly requires a broad IT and business effort, including alternate sites for protection from a localized failure. The application environment is in constant change access depends on many different components that require constant maintenance and change. Recurring changes include: Firmware updates OS patches Capacity upgrades Preventive maintenance on storage hardware Driver updates This constant change, while necessary, introduces the possibility of instability. 5

5 The Exchange environment creates the additional challenge of configuring, adding, and expanding storage to accommodate storage growth. Administrators need to expand Exchange transaction logs periodically, before they fill up and stop the database. Traditionally, these regular maintenance activities must occur during planned system downtime. In today s nonstop environments, administrators need a way to handle ongoing maintenance and growth without shutting out the users. Backup and recovery isn t enough Backup efforts are hampered by short backup windows with growing amounts of data. But if backup times are a problem, they re nothing compared to the worry of long recovery times. Most SLAs define a specific time period for recovering from a problem. A straightforward database recovery from tape can take hours. Administrators need to be able to restore service, and data, quickly in the case of problems. The need for a unified, comprehensive approach to Exchange availability Clearly businesses need more than Exchange administrators with a superior understanding of Microsoft Exchange and a healthy dose of good luck. Exchange administrators need to identify, understand, and respond to a large range of problems that could disrupt access, even if the source of the problem lies outside their expertise. They need to monitor and react to potential outages automatically, according to welldefined response policies. IT organizations need to proactively maintain, upgrade, and otherwise manage the various IT infrastructure components that contribute to service delivery, including server operating systems, network components, and storage systems. The business as a whole needs to be able to keep functioning and communicating in the face of a disaster with cost-effective and flexible disaster recovery planning that doesn t cripple IT budgets. Symantec addresses these needs with a comprehensive, integrated solution for Exchange availability that addresses everything from storage availability to regional disaster recovery. 6

6 The Symantec solution for ensuring Exchange availability Symantec offers an end-to-end, fully integrated solution for maintaining highly available Exchange environments. The Symantec solution reduces planned and unplanned downtime, simplifies the administration of complex environments with a single interface, and offers policy-based failover for disaster recovery/business continuity purposes. Veritas Storage Foundation HA for Windows by Symantec combines Symantec s industryleading volume management technologies (Veritas Storage Foundation for Windows) with flexible, policy-driven failover and application-specific agents (Veritas Cluster Server from Symantec). It expands and enhances the volume management capabilities of Windows servers. Administrators can install the product once, then select among a variety of different options that address specific availability and recovery concerns, such as global failover or online snapshots for quick recovery. The Veritas Storage Foundation HA for Windows solution works with any stand-alone Exchange configuration. Adding Veritas local clustering to the Exchange environment reduces risk and complexity. You can install or back out the clustering software without reinstalling or reconfiguring Exchange. Storage Foundation HA for Windows offers a modular approach to addressing the wide range of potential threats to availability. Organizations can implement the different components in a phased approach or according to their specific needs. Solution components Veritas Storage Foundation HA for Windows from Symantec integrates the following software components: Veritas Storage Foundation by Symantec extends and enhances Windows with the industry s leading volume management technology. Storage Foundation for Windows makes it possible to configure, share, and manage storage for optimal performance and availability, creating a scalable foundation for storage growth. Veritas Cluster Server by Symantec eliminates planned and unplanned downtime by clustering critical applications and the resources they require. Application-specific agents, including an agent for Exchange, monitor and manage the critical components of the IT environment to ensure maximum application availability. Storage Foundation HA for Windows offers the following options, each of which may be activated by a license key on an as-needed basis. Veritas Storage Foundation Dynamic Multi-pathing Option adds fault tolerance to disk storage by using multiple paths between a computer and individual disks. 7

7 Veritas Storage Foundation FlashSnap Option makes it possible to create point-in-time snapshots of storage volumes. These snapshots can be used for snapshot-assisted backups and quick recovery from logical errors. FlashSnap snapshots are fully integrated with Windows Server 2003 Volume Shadow Copy Service (VSS). FlashSnap provides both VSS Provider and VSS Requester support built in to allow creation of Microsoft supported and approved snapshots. The FlashSnap VSS support works with most leading storage arrays, delivering a powerful and flexible solution that eliminates hardware and application restrictions. Global Cluster Option monitors and controls multiple, geographically distributed Veritas clusters as well as Veritas or third-party replication. The Global Cluster Option lets administrators migrate entire data centers with a centralized interface, enabling companies to survive serious local disruptions without significant interruptions to critical services. Veritas Volume Replicator Option by Symantec delivers reliable, storage-independent replication over IP networks for rapid disaster recovery capabilities. The product replicates data at the logical volume level and ensures data integrity and reliability during replication. Complete descriptions of these products can be found at the Symantec Web site ( Using Veritas Storage Foundation HA for Windows, it is possible to create a highly available, resilient storage environment by: Creating storage that automatically expands to meet growing data needs (such as a storage volume for a transaction log) Designing storage configurations that use mirroring or mirroring/striping combinations to protect from the loss of a single disk Identifying and addressing storage hotspots that slow overall application performance Creating point-in-time images for quick recovery from logical errors or data corruption By implementing local or campus clustering with Veritas Storage Foundation HA for Windows, it is possible to: Create a single interface for starting, stopping, and monitoring Exchange Perform maintenance and testing by proactively moving application services to alternate servers in the cluster Meet SLAs by automatically monitoring application delivery and failing over to alternate resources, according to business policies, when problems do occur 8

8 To protect Exchange from site-wide disasters, it is possible to add the Global Cluster and Volume Replicator options to create a disaster recovery environment that enables the fast failover of the entire Exchange environment (or an entire data center). Because the secondary disaster recovery site does not need to match the primary site exactly and can be used for other purposes, companies can leverage DR investments and control the costs of supporting disaster recovery by: Using lower-cost or lower-capacity storage at the off-site recovery location Using a single data center as an off-site recovery location for multiple other data center locations Using the recovery site to run non-critical services, such as development and testing, that can be interrupted, stopped, and replaced with critical applications in case of a global failover The following sections discuss how to use Storage Foundation HA for Windows to specifically address different types of threats to Exchange availability. Creating resilient storage for Exchange The Exchange administrator s first challenge is to create a resilient storage environment for the Exchange data. From the storage perspective, the risks to availability include the following: Failure of a storage component (disk controller, drive, etc.) The failure of a data path (for networked storage configurations) Application outage because an Exchange database or transaction log runs out of room Veritas Storage Foundation HA for Windows protects Exchange users from all of these risks. Veritas Storage Foundation for Windows extends the native data management capabilities, known as Logical Disk Manager (LDM), of Windows 2000 and Windows Server The resulting logical disk/volume capabilities provide the basis for a highly scalable storage environment. Veritas Storage Foundation for Windows supports Fibre Channel and iscsi SAN-based storage, multiple disk groups, enhanced performance tuning, quick data recovery, and drag-anddrop storage management. In addition, the dynamic disk technology provides the framework for disaster recovery solutions using FlashSnap or Volume Replicator. 9

9 Creating highly available storage configurations Using Veritas Storage Foundation for Windows, it is possible to create storage configurations with performance and availability characteristics that surpass the capabilities of the individual disk devices within them. For example, mirrored or RAID 5 volumes can be created from available disks, protecting data from the loss of any single disk. Or it is possible to mirror across multiple disk arrays. If an organization is using a storage area network (SAN), the Storage Foundation cluster disk groups simplify the process of moving storage between servers, while protecting the integrity of the data by ensuring that only one system can use the disk group at any time. In a SAN environment, the Dynamic Multi-pathing Option (DMP) improves availability and performance by providing multiple paths to SAN storage, with I/O load balancing between redundant paths. DMP supports a wide range of storage arrays. Managing growth, nondisruptively Rapid storage growth is a fact of life for most Exchange implementations. The administrator must track and manage that growth, or risk running out of space. In many cases, it s safer to radically overprovision storage than to risk running out of space. Veritas Storage Foundation for Windows offers another alternative. Veritas Storage Foundation for Windows is designed for enterprise storage needs, with the ability to support up to 256 disks in a volume and multiple disk groups per server. The product addresses the challenges of growing storage environments by: Automatically monitoring storage volumes and capacity Sending alerts when storage reaches critical levels Automatically resizing volumes based on predefined policies (thresholds, etc.) Resizing volumes without disrupting storage access For example, Veritas Storage Foundation for Windows can automatically expand the volume containing Exchange mailboxes when it reaches a critical threshold, sending an alert at the same time. When configuring the volume, it is possible to also list a maximum growth size to protect the storage from runaway expansion due to a virus or worm (see Figure 1). 10

10 Figure 1. Veritas Storage Foundation for Windows eliminates running out of space as a potential source of downtime. Shown here is the GUI that allows thresholds to be set by IT. Recovering quickly from logical errors It is infeasible to (good) protect Exchange data from all sources of logical errors. Data corruption, user errors, and viruses all present risks that are nearly impossible to eliminate. The best defense is to undo the effect of these errors quickly and effectively, with minimal data loss. Traditional backup and recovery operations, while essential, aren t up to the task. There s simply too much data involved (data can only be recovered to the point of the last backup) and not enough time for recovery. Veritas Storage Foundation for Windows offers another alternative: point-in-time snapshots of Exchange databases and transaction log files using the FlashSnap Option. A FlashSnap snapshot is an independently addressable volume that mirrors the production volumes. By splitting the mirror, the FlashSnap Option creates a point-in-time image of the data. This image can remain on the production server or be moved to an alternate location. These snapshots have multiple uses. The snapshot itself can be used as a backup source, offloading the backup process from the production data/server. Or multiple snapshots can be used as quick recovery images. The FlashSnap snapshots operate on any storage hardware. The FlashSnap Option uses a FastResynch technology for resynchronizing the split mirror, which reduces the time and CPU impact of rejoining the mirrors. 11

11 In an Exchange environment, restoring the snapshot is just the first part of the process. To reduce data loss, it is necessary to roll forward the transaction log from the point of the image to the point immediately before the failure. Veritas Storage Foundation for Windows automates the quick recovery process, resynching the production image with the most recent FlashSnap image and then restoring transactions. The result is a much faster recovery from data corruption or logical errors than a tape restore. 2:00 10:00 12:00 14:00 16:00 18:00 Full Database Backup with Microsoft Exchange Server Transactions Full Database Backup 13:00 Tape Restore Back in Production Log Roll Forward Transactions Hot snapshot on disk Full Database Backup 13:00 Log Roll Forward In Production (much quicker) FlashSnap Snapshot Restore using Exchange Logs 2:00 10:00 12:00 14:00 16:00 18:00 Figure 2. Recovering Exchange using Transaction logs from a FlashSnap snapshot 12

12 Clustering Exchange servers for protection from component failures Organizations can protect Exchange environments from a wide range of component failures by implementing local clustering for availability. Storage Foundation HA for Windows integrates Veritas Cluster Server technology, which provides highly flexible, scalable failover clustering with workload management capabilities. In a Veritas Cluster Server cluster, multiple servers are linked with shared storage and private Ethernet heartbeats. Each system in the cluster can access the storage of any other system. Applicationspecific agents monitor the various components of the application environment for failures. Although it s possible to share storage with switched SCSI devices, most clusters implement SAN-based storage to support a larger number of nodes. Veritas Cluster Server implements a policy-based response to failures, attempting a restart and/or automatically moving a failed application to another resource in the cluster. Veritas Cluster Server has a built-in sense of cluster load and can relocate application components based on available capacity augmented with a rules-based policy engine. Veritas Cluster Server offers a highly scalable, flexible alternative to Microsoft s native clustering capabilities, and includes: Support for up to 32 nodes and 4 Global Cluster sites Policy-based failover Workload management Automated DNS updates Support for failover across subnets Support for mount points The Veritas Cluster Server technology imposes no restrictions on the Exchange configuration, such as a maximum number of mailboxes. Any stand-alone Microsoft Exchange implementation will run in a Storage Foundation HA for Windows cluster, without being reconfigured or rebuilt. And Storage Foundation HA for Windows clustering provides the groundwork for campus or disaster recovery clustering solutions, as described later in this paper. 13

13 Cluster configurations Veritas Storage Foundation HA for Windows supports either Active/Passive or any-to-any cluster configurations for Microsoft Exchange. In an Active/Passive cluster, one spare system serves as the standby or passive system for a production server. In an any-to-any cluster, Microsoft Exchange can fail over to any server in the node that is not already running Exchange and that has adequate capacity: Database A with Microsoft Exchange Server Database A with Microsoft Exchange Server Database C Veritas Cluster Server Storage Foundation Veritas Cluster Server Storage Foundation! Veritas Cluster VCS Server Storage Foundation Database C with Microsoft Exchange Server Veritas Cluster Server Storage Foundation Figure 3. Advanced failover logic initiates an Exchange database fails over to another server not running Exchange The cluster can include numerous other servers running different applications, consolidating management of high availability environments and providing spare capacity for critical applications. Veritas Cluster Server clusters can support up to 32 nodes. 14

14 Service groups Veritas Cluster Server encapsulates the resources required for each application into a service group creating a virtualized application that can be moved between the nodes of the cluster. Each Exchange service group contains a set of dependent resources the lower-level components that an application requires to operate successfully. Resources include disk groups, disk volumes, file systems, IP addresses, and NIC cards. Figure 4. A Veritas Cluster Server dependency graph that shows the ordered sequence of resources as they start to come online when a failover occurs. Once the cluster is configured, administrators can define failover policies for critical applications. Veritas Cluster Server supports flexible, customized failover policies that combine rules with workload management. Policies are configured using a graphical interface or command line interface, and may be supplemented with custom scripts if required. Veritas Storage Foundation HA for Windows also offers a simulator for testing failover. For example, policies might specify which applications have failover priority. Veritas Storage Foundation HA for Windows provides the opportunity to test failover so that failover occurs as expected, before an actual crisis occurs. 15

15 Local clustering with Storage Foundation HA for Windows addresses many of the challenges of Exchange administration: The software automatically monitors all of the application components and responds appropriately in case of a problem, failing over to other resources if necessary. Administrators can proactively switch applications to other resources to perform routine maintenance or upgrades on application components, such as server upgrades or OS patch applications. Protecting Exchange from localized failures with campus clusters Local clusters are, by definition, local. Although they can survive the failure of individual components, they cannot protect you from the loss of an entire site. A power system failure or fire at the data center would disrupt service. Many large organizations enhance recoverability by stretching the clustering across different facilities that are nearby but in the same general area. These campus clusters or metropolitan clusters combine high availability with an intermediate level of disaster recovery. They can protect against many localized disasters, but not against major area-wide disturbances such as hurricanes and earthquakes. The limiting factor in the campus cluster is the distance between the nodes, which must maintain heartbeat communications. For Veritas Storage Foundation HA for Windows, the latency in the node interconnect is the key factor. Symantec recommends that network latency not exceed 500 milliseconds between nodes. Any further, and the better solution is a global cluster of clusters communicating over wide area networks. (This approach is described in the next section.) The data is mirrored across the cluster nodes using a SAN infrastructure. For example, a single cluster could span two distinct sites, with a SAN infrastructure in place between them. The Exchange administrator uses Veritas Storage Foundation HA for Windows to mirror the Exchange data volumes automatically between the two locations. 16

16 Local Clustering (LAN) Metro DR with Remote Mirroring Metropolitan Disaster Recovery (MAN) Metro DR with Replication Wide Area Disaster Recovery (WAN) Storage Foundation HA for Windows + FlashSnap + Veritas Volume Replicator Option + Disaster Recovery Option (Veritas Volume Replicator Option + Global Cluster Option) Figure 5. Metropolitan area clustering using volume mirroring A financial company in Manhattan might maintain fibre connections to a data center in New Jersey, and mirror data between the two sites. In the event of a failure at the Manhattan facility, the application automatically switches over to the New Jersey site and service is restored within minutes, with no data loss. This configuration works with Veritas Storage Foundation for Windows volume-based mirroring or a third-party array that provides data mirroring. Although the network infrastructure is needed to support the communications between servers and between mirrored volumes, campus clustering with mirroring requires no additional software licenses beyond those available with Veritas Storage Foundation HA for Windows. If no SAN infrastructure exists, organizations can consider using volume replication to replicate data between nodes in a replicated data cluster. Protecting Exchange with replicated data clusters For organizations that do not have a SAN infrastructure in place, it s also possible to create a campus cluster using data replication either volume replication through Veritas Storage Foundation HA for Windows or replication provided by a third-party array. 17

17 Local Clustering (LAN) Metro DR with Remote Mirroring Metropolitan Disaster Recovery (MAN) Metro DR with Replication Wide Area Disaster Recovery (WAN) Storage Foundation HA for Windows + FlashSnap + Veritas Volume Replicator Option + Disaster Recovery Option (Veritas Volume Replicator Option + Global Cluster Option) Figure 6. Metropolitan area clustering using IP-based volume replication The data is replicated across sites, which eliminates the need for the SAN infrastructure across sites. Synchronous replication provides the best data protection, although Veritas Volume Replicator provides a capability that switches to asynchronous replication when network latency is a problem. For organizations with primary data centers in lower-risk areas (without the threats of widespread problems from earthquakes or hurricanes), maintaining a disaster recovery facility in close proximity to the primary site is an easy and cost-effective approach to disaster recovery. Most disasters are localized, so campus clusters offer rapid disaster recovery for a wide range of potential problems. If a large geographic separation between sites is required, then a wide area clustering solution is preferable. Disaster recovery with global clustering For business continuity in the face of a broader disruption, the best protection is the ability to resume operations, nearly instantly and without data loss, at an alternate site located a significant distance from the primary site. Using Veritas Storage Foundation HA for Windows with the Global Cluster and Volume Replicator options, it is possible to replicate data between two separated sites, and switch application services between them with a single mouse click. 18

18 Local Clustering (LAN) Metro DR with Remote Mirroring Metropolitan Disaster Recovery (MAN) Metro DR with Replication Wide Area Disaster Recovery (WAN) Storage Foundation HA for Windows + FlashSnap + Veritas Volume Replicator Option + Disaster Recovery Option (Veritas Volume Replicator Option + Global Cluster Option) Figure 7. Wide area clustering with replication across a WAN and using a Veritas Global Cluster Option Site Migration (VCS with GCO) Site A Failover (VVR) WAN Site B Secondary Primary Replicated Volume Group Replication (VVR) Log Replicated Volume Group Figure 8. Site failover using Veritas Cluster Server with Global Cluster Option and Veritas Volume Replicator Option Global Cluster Option creates a cluster of clusters. The local cluster remains at one location. Should an application fail, the clustering software will first try to restart it within its local cluster. If the local cluster fails, then the Veritas Cluster Server software notifies the administrator, who can verify the problem and move the application with the touch of a single button. Failover and replication can occur over any distance. 19

19 It is possible to replicate the data between sites using either the Volume Replicator Option (which offers hardware-independent wide area replication over IP) or third-party hardware-based replication. Primary and secondary sites Global clustering depends on the existence of an alternate site. The cost of the alternate site is a major concern for many organizations how many can afford to have a duplicate data center, somewhere distant from the primary data center, ready to accept production processing at a moment s notice in case of a disaster? Using Veritas Storage Foundation HA for Windows helps organizations manage alternate sites without damaging the IT budget, offering: Server savings: The secondary site doesn t need to be identical in computing resources to the primary site. The DR site might have less capacity, but still can provide enough computing resources to run Exchange should a disaster strike. Storage savings: The secondary site can use the same storage from a single vendor, different storage from a single vendor, or storage from multiple vendors, as long as the overall capacity is sufficient. Veritas Storage Foundation for Windows and the Volume Replicator Option do not require the same disk array on both sites of the replication link, and work effectively in a heterogeneous storage environment. Alternate uses for the DR environment: Many organizations actively use the DR site for other purposes that could be stopped in the case of a true disaster. For example, the secondary site could host development and QA processes. The next step would be to configure, as part of the failover or switchover process, the graceful shutdown of development processes before bringing the production instance online. Flexible global configurations: In a company with multiple data center locations, one site could serve as the DR site for multiple other data centers. In many cases, it is possible to use existing resources to achieve a level of protection not previously available at a manageable cost. 20

20 Veritas Volume Replicator Option To support application migration, the production data must be resident and up-to-date at the secondary site. It s possible to use hardware-based replication, particularly if the solution is already in-house. The Volume Replicator Option makes it possible to consistently and reliably replicate data across IP networks using any available storage. The Storage Foundation Volume Replicator Option replicates the contents of each volume across a wide area network to the secondary site. It is completely transparent to the application components. Unlike traditional block-based approaches, Volume Replicator replicates I/O instead of disk tracks to ensure that the data is always replicated in a consistent fashion. Volume Replicator supports both synchronous and asynchronous replication. Using synchronous replication, the initial write is not committed until the data has been replicated successfully. This guarantees that there will be no data loss in case of a site failure, but it impacts application performance because the application at the primary site waits for the transaction to travel to the secondary site and back before the transaction is committed at the primary site. Over long distances, synchronous replication may introduce unacceptable write delays in production systems. Using asynchronous replication, Volume Replicator commits the data at the primary site immediately and then queues replication operations for network availability. Asynchronous replication does not impact the application performance at the primary site, but there may be some potential for data loss. Typical data loss between sites over long distance can be measured in milliseconds. Volume Replicator is unique in the market because it enforces write order on the replicated site, ensuring data integrity and consistency, thereby guaranteeing that the data will be recoverable at the secondary site. Most sites select asynchronous replication for long-distance, global failover scenarios. Volume Replicator can dynamically switch to asynchronous replication if network latency is a problem during synchronous replication. 21

21 Scenario: Site migration To understand how the integrated Symantec solution can migrate the Exchange environment across sites for disaster recovery, it s worth walking through the failover scenario. Consider the case of a company with production facility running Exchange servers in Dallas and a DR/failover site in Charlotte. The Charlotte site could run other services, which would then be shut down in case of a failover. For the purposes of this example, we ll only look at the Exchange application. 2. Reconfigure network Site A Failover (VVR) WAN Site B 4. Restart services, Exchange Secondary Primary Replicated Volume Group Replication (VVR) 1. Stop replication Replicated Volume Group 3. Promote storage Figure 9. Four-step process that initiates a site failover and restarts Exchange services at the secondary site. If the Dallas site has a major problem, such as a power failure or flooding, an operator would be notified that the site in Dallas was down. The operator would then log on to a Web GUI from anywhere in the world to verify that the site is actually not available. At that point the operator would initiate the failover with a single click to fail over Exchange services. 22

22 The solution would automatically perform the following steps: 1. Stop replication: In a failover, the first thing that occurs is that replication between the primary and secondary sites stops. (If the primary site has experienced a dramatic failure, then replication has already stopped.) 2. Reconfigure the network: The Global Cluster Option reconfigures the logical network, migrating the virtual IP address for the application to the Charlotte site s systems. 3. Promote the replicated data: The Global Cluster Option promotes the secondary data to primary status. If a secondary site is still available, Volume Replicator can replicate to another site to maintain data availability. 4. Restart services: The Global Cluster Option starts the Cluster Server service groups, bringing Exchange back online. At this point, Exchange is now running successfully at Charlotte typically in a matter of minutes. Users would access without being aware that the application was running in a secondary site. When the disaster has passed and the primary data center is again ready, the operator can update the Dallas storage, restart replication, and then switch the application back to its original location, using a central console. The only data sent back to the Dallas site is the data that changed while operations were occurring in Charlotte; a complete re-initialization is not necessary. Compare this to the processes companies manage today. First, companies maintain the DR site at the same update and patch levels as the primary site difficult when the site is physically removed from the primary data center. Coordinating change control for multiple sites for each production application presents a significant challenge. Then, when it s time to switch an application to the DR site, IT has to move tapes and staff between facilities, rebuild servers, and load operating systems and backup software while manually tracking and maintaining what s available at each site and relying on experts to bring up the environment at the DR site appropriately and consistently. The process can be very time-consuming and expensive, as well as being subject to possible human error, during a complete site outage. The Symantec solution eliminates the change control problems, as all changes to the primary application (and its subsidiary components) are replicated to the DR site automatically. The failover is automated and nearly instantaneous, reducing risks while improving application availability. 23

23 Conclusion Veritas Storage Foundation HA for Windows offers organizations that depend on Microsoft Exchange a single solution for protecting availability from a wide variety of problems, ranging from storage-caused outages to server failures and site-wide disruptions. The Symantec solution is Microsoft-certified and well integrated in the Windows environment. Microsoft and Symantec have worked together for many years to improve storage manageability on the Windows platform. Microsoft selected Veritas to develop the disk management software for both Windows 2000 and Windows Server The Veritas Storage Foundation for Windows solution further builds on the dynamic disk and dynamic volume capabilities now native to the Windows platform ensuring customers receive a more robust and time tested Microsoft compliant volume management platform for their storage management needs. Whether the greater concern is data corruption or business continuity, Veritas Storage Foundation HA for Windows provides a packaged, integrated solution tailored to the Exchange environment, with the potential for expanding the solution through integrated agents and options. For additional information about Symantec, its products, or the location of an office near you, please call our corporate headquarters or visit our Web site at Useful Links: Storage Foundation for Windows Symantec Enteprise Products 24

24 About Symantec Symantec is the world leader in providing solutions to help individuals and enterprises assure the security, availability, and integrity of their information. Headquartered in Cupertino, Calif., Symantec has operations in more than 40 countries. More information is available at For specific country offices and contact numbers, please visit our Web site. For product information in the U.S., call toll-free 1 (800) Symantec Corporation World Headquarters Stevens Creek Boulevard Cupertino, CA USA +1 (408) (800) Copyright 2006 Symantec Corporation. All rights reserved. Symantec, the Symantec logo, FlashSnap, Veritas, and Veritas Storage Foundation are trademarks or registered trademarks of Symantec Corporation or its affiliates in the U.S. and other countries. Microsoft, Windows, and Windows Server are either registered trademarks or trademarks of Microsoft Corporation in the United States and/or other countries. Other names may be trademarks of their respective owners. Printed in the U.S.A. 09/