Advisor Assessment Report

Transcription

1 Advisor Assessment Report Client: Reference Systems Technology Incorporated Project: Reference Project Model: 2014 Reference Model - Unix/Linux Report: Virtustream Branding Sample Report Version: 1.0 Date: Thursday, January 23, 2014 Prepared By: Derek Bambach Prepared For: internal review Classification: Discussion Document CC: Reference Systems Technology Incorporated Virtustream Advisor Assessment Report v

2 Virtustream, Inc. North America Bethesda, Maryland (Headquarters) 4800 Montgomery Lane, Suite 1100, Bethesda, MD Tel: Fax: San Francisco, California 455 Market Street, Suite 1250, San Francisco, CA Tel: Fax: Atlanta, Georgia 6 Concourse Parkway, Suite 1930, Atlanta, GA Tel: Fax: European Offices London, UK 3 rd Floor, Leadenhall Street, London, EC3A 3DH Tel: +44 (0) Fax: +44 (0) Kaunas, Lithuania Coming Soon Copyright 2014 by Virtustream, Inc. All rights reserved worldwide. Reference Systems Technology Incorporated Virtustream Advisor Assessment Report

3 Document Control: The following document details the effect that Cloud and Virtualization would have on the future Information Technology strategy of Reference Systems Technology Incorporated. It is based on the information provided to Virtustream in a series of meetings and also any performance analysis that is available at the time of writing. Information contained in this document is accurate to the best of Virtustream knowledge at the time of publication and is required to be treated as confidential. It should not be reproduced or made available in any form to persons outside the group directly responsible for evaluating its contents Any performance data provided is based on relevant information made available to us at the present time, and must (except where stated otherwise) be regarded as an estimate only, since the actual performance and functionality of any computer system will depend upon a variety of factors, not all of which are related to the products and services that may ultimately be supplied. For information purposes, please find below the document control details: Client: Reference Systems Technology Incorporated Project Name: Reference Project Model Name: 2014 Reference Model - Unix/Linux Report Name: Virtustream Branding Sample Report Doc Version: 1.0 Report Date: Thursday, January 23, 2014 Prepared By: Derek Bambach Prepared For: internal review Classification: Discussion Document CC: Reference Systems Technology Incorporated Virtustream Advisor Assessment Report

4 Contents 1. Advisor Study Objectives... 3 Overview of Reference Systems Inc...3 Project Background...3 Reference Systems Inc. Objectives for the Cloud...3 Cloud Options for Consideration...3 Concerns about the Cloud...3 Success Criteria for Reference Systems Inc Executive Summary... 4 Overview...4 Summary of Resource Utilization...4 Main Areas of Service Improvement...4 Conclusions and Recommendations General IT Profile... 6 IT at Reference Systems Inc. (Environment Summary)...6 Current Data Center Information...6 Server Estate Information...6 Device Estate Information...6 Network and Access Overview Business Continuity, Disaster Recovery, Backup and Restore... 7 Overview...7 Advisor Interview Results...7 Virtualization and Cloud Services for Back-up and Restore...7 Virtualization as an Enabling Technology for Business Continuity...7 Disaster Recovery Objectives...8 Business Continuity Requirements...8 Higher Systems Availability...9 Business Continuity Conclusions and Recommendations Security and Compliance Requirements Overview Security Audit Requirements Security and Compliance Recommendations Performance Statistic Summaries Overview Measured Candidates CPU Peak Hour Demand Load (56 candidates) CPU Consumption (56 candidates) Memory Consumption (56 candidates) Disk Activity: Input/Output Operations (56 candidates) Disk Activity: Volume (56 candidates) Network Data Volume (55 candidates) [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 1

5 Storage Capacities (56 candidates) Aggregate Resource Utilization over Time Aggregate CPU Utilization Aggregate Memory Utilization Aggregate Disk IOPS (Read) Aggregate Disk IOPS (Write) Aggregate Disk Volume (Read) Aggregate Disk Volume (Write) Aggregate Disk IOPS (Total) Aggregate Disk Volume (Total) Aggregate Storage Utilization Aggregate Network Utilization Week by Week Performance Analysis for Virtualization Overview Measured Performance Data Period Measured Performance Data Period Measured Performance Data Period Measured Performance Data Period Resource Utilization Analysis Overview µvm Consumption (56 candidates) Potential Impact of Virtualization Proposed Virtualization Host Resources Combined Most Intensive Resource Utilization modelling Potential Impact of Cloud Infrastructure Proposed Cloud µvm Resources Current Overall Resource Utilization Environmental Impact Overview Existing Infrastructure Power Consumption Impact of Virtualization Solution Impact of Cloud Solution Report Signoff Target List About Virtustream, Inc Appendix - Virtualization Explained Appendix - Cloud Explained [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 2

6 1. Advisor Study Objectives Overview of Reference Systems Inc. [Questionnaire Section 01 - Question 03] Reference Systems Inc. provides real time system process monitoring for industrial, utility, and military markets worldwide. The company provides tools to monitor and report on realtime conditions and changes to manufacturing, utility, and infrastructure support systems to insure system integrity and to create an automated alert and response system. Reference Systems Inc. offers its monitoring tools directly and through third party distributors, and also offers additional out-sourced monitoring services to users of its monitoring tools. Project Background [Questionnaire Section 01 - Question 03] Reference Systems Inc. provides real time system process monitoring for industrial, utility, and military markets worldwide. The company provides tools to monitor and report on realtime conditions and changes to manufacturing, utility, and infrastructure support systems to insure system integrity and to create an automated alert and response system. Reference Systems Inc. offers its monitoring tools directly and through third party distributors, and also offers additional out-sourced monitoring services to users of its monitoring tools. Reference Systems Inc. Objectives for the Cloud [Questionnaire Section 34 - Question 03] Summary of the Client Objectives for the Cloud. Cloud Options for Consideration [Questionnaire Section 34 - Question 04] Summary of the Cloud Options under Consideration for this assessment. Concerns about the Cloud [Questionnaire Section 34 - Question 05] Summary of the Client's concerns about the Cloud. Success Criteria for Reference Systems Inc. [Questionnaire Section 34 - Question 06] Consumption model, right class of server, agility, replication, DR, backup, Increase Availability, Risk Mitigation, cost effectiveness. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 3

7 2. Executive Summary Overview [Questionnaire Section 34 - Question 02] Executive Overview summary based on Client interviews. This report discusses the opportunity that cloud services and virtualization technologies present to Reference Systems Inc. and the approaches required to implement and complete a successful transformation from its current status. Summary of Resource Utilization Memory, CPU, IOPS [Questionnaire Section 34 - Question 07] Summary of current Resource Utilization. The below chart summarizes the usage results of Reference Systems Inc. s infrastructure in aggregate. Detailed analysis of the systems and the implications for virtualization and cloud services is provided in Chapters 6-9. Resource Total during Peak Period % Utilized µvms CPU (Mhz) 366, % 1,833 Memory (MB) 1,920, % 2,687 Disk I/OPS 1,826,416 N/A 1,274 Network (B/s) 1,767,340 N/A 4,307 Main Areas of Service Improvement Potential Efficiencies from Virtualization and Cloud Services Industry averages place CPU utilization lower than 15% per server Reference Systems Inc. is averaging 18.0% at peak. Virtualization consolidates all this inefficiency onto fewer platforms to drive up utilization without compromising performance / reliability. Total Cost of Ownership Systems Availability Business Continuity [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 4

8 Conclusions and Recommendations [Questionnaire Section 34 - Question 20] Conclusions and Recommendations based on Client interviews and conversations. The Advisor Report lays the initial groundwork on the current configuration of Reference Systems Inc. s IT infrastructure, and provides a 4-week profile of the system usage. Virtustream recommends that Reference Systems Inc. use this Advisor Report as the baseline for developing a Cloud Strategy and Implementation Project. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 5

9 3. General IT Profile IT at Reference Systems Inc. (Environment Summary) [Questionnaire Section 34 - Question 09] Summary of the Client's IT environment based on Client interviews. Current Data Center Information Reference Systems Inc. currently operates servers at main locations. Location Details: [Questionnaire Section 12 - Question 08] New York, Los Angeles. Server Estate Information The existing (in scope) server estate is spread between these data centers, and this comprises approximately 56 servers, of which 70 servers are operating in a virtual environment. The replacement program for the servers is every 4 years and approximately of the servers are replaced each year; this represents approximately 0 servers per year. The growth rate on the server estate is estimated at year on year, therefore this represents an additional servers for the next year. Multiple Production environments, related test and development environments Device Estate Information The client access device deployed at Reference Systems Inc. is predominantly a typical full function workstation device of which there are approximately in use. There is also some use of Microsoft Terminal Services via the Remote Desktop Protocol (RDP) for about 50 users with approximately 2000 laptops. Network and Access Overview The following diagrams provide the network and access structure at Reference Systems Inc. [ADD SUPPORTING DIAGRAMS HERE] [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 6

10 4. Business Continuity, Disaster Recovery, Backup and Restore Overview Much of the content of this report focuses on new technologies and different methods to achieve objectives via virtualization. However, one area that is still as valid as ever is the requirement for a Business Continuity Strategy that includes a robust and reliable backup and recovery solution. Virtustream has extensive experience in the specification, implementation and support of Enterprise backup solutions, and virtualization has until recently presented a significant challenge with regard to backup and recovery strategies. Advisor Interview Results [Questionnaire Section 34 - Question 13] Overview based on the Advisor Interview results. For further detail on the results of this area, please see the Business Continuity, Back-up, and Restore section of the Advisor Questionnaire Summary. Virtualization and Cloud Services for Back-up and Restore Virtualization as an Enabling Technology for Business Continuity Reference Systems Inc. has a separate strategic review being conducted on how to improve the current disaster recovery capabilities, and virtualization has been recognized as a potential enabling technology that could assist in the delivery of such a strategy. The assumption that virtualization could potentially have benefits in assisting such a strategy is well founded as virtualization is the cornerstone technology to enabling organizations to simplify and automate the tasks required to provide reliable and rapid disaster recovery. If all the requisite components in the technology stack illustrated below are implemented, then this recovery can begin to move towards business continuity as in a solution that does not need recovery as it has avoided the outage in the first place. This report uses the following terminology to explain the evolution of disaster recovery capabilities. Tier Terminology Explanation 1 Business Continuity 2 Disaster Recovery 3 High Availability The ability to cope with most types of outage and continue to provide key services with no or very limited disruption and little or no restoration services In the event of an interruption to services a plan is activated to recover the systems within the agreed timelines. Services are interrupted until recovery services have been performed. The ability to provide higher availability for systems within the confines of no major outages. Within each of the tiers, it is necessary to implement techniques, designs, technologies and processes to ensure that the objective can be met. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 7

11 Disaster Recovery Objectives Business Continuity Requirements Reference Systems Inc. will need to consider the following components to enable a transition to a continuity solution. The table below indicates the required stack of components to deliver true business continuity Component Status Requirement Multi-Site Data Center Resilient Communications Remote access Replicated Data Service availability / resource provisioning Operational planning [Add status here] [Add status here] [Add status here] [Add status here] [Add status here] [Add status here] To be able to function in the event of an outage it is a necessity that operations can operate from more than one location. The solution is only as strong as its weakest link and therefore the communications backbone needs to be multi-homed and routed. The ability to utilize applications remotely is critical as the data is only useful if it can be accessed. The requirement to replicate data between sites so that the need to restore from backups is minimized or removed totally. Business continuity is all about the ability to remove lead times. This forms the mainstay of this report to be able to measure and quantify the resource required to operate the services in a virtualized presentation. Without virtualization duplicate hardware for all services that are deemed required in a disaster scenario would be required. The requisite planning and testing of a Business Continuity Solution, the creation of the procedures to be adhered to and the test of their success. A sound business continuity plan is an essential part of the process. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 8

12 Component Status Requirement Semi-Automation [Add status here] To enable the most rapid restoration of services in the event of an outage it is a requirement that automation or semi-automation of processes be implemented to remove the lag introduced and the risk of human error to repeat the tested in a true outage situation. It is also the key to testing the business continuity plan. The functionality provided by the semi-automation solution is covered in detail in the operational management section of this report. In summary by combining the functionality of VMware Site Recovery Manager and its other automation tools it will be possible for Reference Systems Inc. to implement the semi-automation frameworks required to simplify and more consistently enable site failover and fail-back. Higher Systems Availability Like any organization, Reference Systems Inc. has expressed a desire to have enhanced systems availability from its services. This is broken down into the following key areas: Ability for systems to cope with outages Improved disaster recovery capabilities through intelligent storage management Improved agility to enable planned systems maintenance Standardization Improved management Service Outages To minimize the ability of system outages to affect services, Reference Systems Inc. should consider a design that addresses two critical areas: storage and resource pools. The resource pools are handled by the High Availability function of VMware vsphere server where the loss of a host system running the underlying VMware ESX Hypervisor would instigate the servers that were being serviced by the downed host being restarted on an alternative host. The host that will operate these servers is decided by the Distributed Resource Scheduler (DRS) function of VI3 which finds the best placed host to service the requests. By building in enough resource to cope with host outages, this can facilitate both an unplanned failure and planned maintenance in working hours. Systems Maintenance Systems Maintenance is essentially to maintain the wellbeing of the IT infrastructure, unfortunately this can lead to planned downtime to avoid unplanned downtime, the opportunity to complete this required task is getting less and less as organizations tolerance to any interruption of service reduces. In line with the Services Outages section, it is necessary to build in capability to perform maintenance to both the storage and resource pools and therefore both solutions need to have this functionality. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 9

13 Standardization A key to maintaining high availability is to maintain approved standards as this minimizes the effect of variations. Virtualization allows standard server templates to be created on a platform of uniform hardware and therefore one of the most disruptive areas of server support that of non-standard builds or hardware is avoided, thus improving availability. Business Continuity Conclusions and Recommendations [Questionnaire Section 34 - Question 01] Summary of conclusions & recommendations regarding Business Continuity. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 10

14 5. Security and Compliance Requirements Overview [Questionnaire Section 34 - Question 15] Summary of the security and compliance overview based on Client interviews. Security Audit Requirements [Questionnaire Section 34 - Question 16] Summary of the Client's Security Audit Requirements based on Client interviews. Security and Compliance Recommendations [Questionnaire Section 34 - Question 17] Summary of our Security and Compliance Recommendation for the Client. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 11

15 6. Performance Statistic Summaries Overview Cloud computing and Virtualization are essentially all about resource management. These approaches are, in some ways, a return of the methodologies of mid-range and mainframe computing models. Therefore to understand the areas that could benefit the most from virtualization, it is important to monitor and asses the current utilization models. The principle benefit of virtualization is in its ability to utilize underlying physical resources in a more intelligent way by sharing resources based on service level policies. It is this ability to be flexible to the varying demands placed upon the consolidated resources that enables organizations to drive efficiencies within the organization. By removing significant dependencies on the underlying physical resources, it is possible to design highly available, performance enhanced and efficient shared platforms. Once operating in such a manner, further enhancements can be made to improve the management and optimization of IT infrastructure through service-level automation frameworks enabling organizations to manage more with less and more effectively. Measured Candidates Performance Characteristics Overall Summaries Each area of critical importance to virtualization/consolidation is summarized complete with utilization graphs. This includes the following areas: CPU and Memory Trends and efficiencies Disk IOPS Trends Disk IOPS data volume trends Networking Trends Storage Capacity Requirements Utilization Detail As part of the process, we have run the Virtustream Advisor to collect inventory and performance information over a four week period. The statistics have been collated to help formulate a server consolidation solution targeted at a cloud architecture. We use the peak load values when considering consolidation, not the weekly average. As statistics are received, they are evaluated to determine what hour of the day the peak load is placed on each individual server. The peak load is determined by evaluating load for a minimum of four weeks. The hour with the consistently high load will be deemed the peak load hour, and its average value becomes the peak load for the server. These peak loads are then correlated across the broader collection of servers to determine the optimal mix of workloads with a virtual or cloud infrastructure. Proposed Consolidation of Resources Using the statistics gathered in the planning exercise, Virtustream is able to estimate actual resources required in the target cloud environment(s) to ensure that Reference Systems Inc. has the necessary compute resources required to meet its demands. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 12

16 Performance Characteristics Week by Week Summaries Each area of critical importance to virtualization/consolidation is summarized to create the week s virtualization candidate profile, and this is modelled against the proposed host resources that form the virtualization platform. It shows the demand that the measured systems would place upon the consolidated design. A Guide to the Measured Statistics Resource CPU Util. MHz RAM Used MB Disk Size GB Disk Used GB Disk IOPS Disk Reads/Writes B/s Network Traffic B/s Explanation A calculation of the peak average MHz load placed upon the existing server estate based on % utilization against the total amount of CPU installed A calculation of the utilized memory (RAM) in MB against the allocated resource The total amount of Hard Disk space allocated to the server The total amount of allocated disk space actually used by the server The total storage Input/output operations The total of the IOPs with regard to read and write operations A measure of how busy the server network card is, taken from the average data throughput in bytes per second Data Collection Periods Period One: Number of Candidates: 56 Period Two: Number of Candidates: 56 Period Three: Number of Candidates: 56 Period Four: Number of Candidates: 56 Applied Resource Model Adjustments* Target Expansion: % [if 100%, no adjustment applied] CPU Performance: % [if 100%, no adjustment applied] *Adjustment values of 100% = NO ADJUSTMENT APPLIED [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 13

17 12 AM 1 AM 2 AM 3 AM 4 AM 5 AM 6 AM 7 AM 8 AM 9 AM 10 AM 11 AM 12 PM 1 PM 2 PM 3 PM 4 PM 5 PM 6 PM 7 PM 8 PM 9 PM 10 PM 11 PM CPU Peak Hour Demand Load (56 candidates) Reference Model - Unix/Linux: Servers at Peak CPU per Hour Commentary Ideally, organizations target an even spread of demand throughout the 24 hour period. If there are concentrated periods of load being experienced this can typically be causes by some of the following activities: 1) Backup schedules 2) Anti-Virus Scans 3) Update periods 4) Logon and Logoff cycles Servers Per Hour at MAX Virtustream models on the load placed at peak average demand. The graph illustrates that there is a peak demand for CPU at a concentrated period between 3:00 AM - 4:00 AM when the highest number of servers (5) experience their most demanding hour. This represents 8.9% of the entire server estate and as a concentrated workload. [Questionnaire Section 35 - Question 02] Additional Commentary - CPU Peak Hour Demand Load. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 14

18 CPU Consumption (56 candidates) CPU Utilization (MHz) 400, , , , , , ,000 50, Reference Model - Unix/Linux: CPU (MHz) 350, , ,538 1/7/2013 1/14/2013 1/21/2013 1/28/ ,543 MIN AVG 95th% MAX Commentary The CPU Consumption graphs show how the demands placed on the systems have varied from week to week - from a low consumption peak of 301,543 MHz observed in week #N/A to the highest peak in week 3 at 366,538 MHz. This equates to an increase of 64,995 MHz across all of the servers, or an average increase of 64,995 MHz per machine for CPU consumption from the lowest to the highest observed peak. [Questionnaire Section 35 - Question 01] Additional Commentary - CPU Consumption. CPU (Percent) 2014 Reference Model - Unix/Linux: CPU (% of Asset) 20% 15% 17% 16% 18% 15% 10% 5% 0% 1/7/2013 1/14/2013 1/21/2013 1/28/2013 CPU MIN CPU AVG CPU 95th% CPU MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 15

19 CPU Utilization Average vs. Maximum (MHz) 2014 Reference Model - Unix/Linux: CPU [AVG + MAX] (MHz) 400, , , , , , , , , , , , /7/2013 1/14/2013 1/21/2013 1/28/2013 AVG MAX CPU Utilization Average vs. Maximum (Percent) 2014 Reference Model - Unix/Linux: CPU [AVG + MAX] (%) 20% 15% 10% 5% 10.4% 17.2% 9.7% 15.8% 10.3% 18.0% 10.1% 14.8% 0% 1/7/2013 1/14/2013 1/21/2013 1/28/2013 CPU AVG CPU MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 16

20 CPU Utilization at Maximum (MHz and Percent Available) 2014 Reference Model - Unix/Linux: MAX (MHz & %) 100% 80% 60% 1,684,920 1,714,735 1,669,243 1,734,238 40% 20% 0% 350, , , ,543 1/7/2013 1/14/2013 1/21/2013 1/28/2013 MAX CPU MAX CPU Utilization at Average and Maximum (Percent Available) Percent used at AVG AVG CPU AVG 10% 90% [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 17

21 Percent used at MAX MAX CPU MAX 16% 84% [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 18

22 Memory Consumption (56 candidates) Memory Utilization (MB) 2014 Reference Model - Unix/Linux: Memory (MB) 3,000,000 2,643,511 2,630,420 2,645,193 2,641,670 2,500,000 2,000,000 1,500,000 1,000, , /7/2013 1/14/2013 1/21/2013 1/28/2013 Commentary The Memory consumption graphs show how the demands placed on the system vary from week to week. The highest period of memory utilization observed in week 3 with 2,645,193 MB of memory consumed at peak; the lowest result was collected in week 2 with a total of 2,630,420 MB utilized, yielding a net difference of MB or 264 MB per server on average. [Questionnaire Section 35 - Question 03] Additional Commentary - Memory Consumption. MIN AVG 95th% MAX Memory Utilization (Percent) 2014 Reference Model - Unix/Linux: Memory (% of Asset) 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 83% 83% 83% 83% 1/7/2013 1/14/2013 1/21/2013 1/28/2013 Memory MIN Memory AVG Memory 95th% Memory MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 19

23 Memory Utilization Average vs. Maximum (MB) 2014 Reference Model - Unix/Linux: Memory [AVG + MAX] (MB) 3,000,000 2,500,000 2,000,000 1,500,000 1,000, , ,643,511 2,630,420 2,535,492 2,645,193 2,517,897 2,559,793 1/7/2013 1/14/2013 1/21/2013 2,641,670 2,487,056 1/28/2013 AVG MAX Memory Utilization Average vs. Maximum (Percent) 2014 Reference Model - Unix/Linux: Memory [AVG + MAX] (%) 100% 80% 60% 83.4% 80.0% 83.0% 79.4% 83.5% 80.8% 78.5% 83.3% 40% 20% 0% 1/7/2013 1/14/2013 1/21/2013 1/28/2013 Memory AVG Memory MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 20

24 Memory Utilization at Maximum (MB and Percent Available) 2014 Reference Model - Unix/Linux: MAX (MB & %) 100% 80% 526, , , ,862 60% 40% 2,643,511 2,630,420 2,645,193 2,641,670 20% 0% 1/7/2013 1/14/2013 1/21/2013 1/28/2013 MAX Memory MAX Memory Utilization at Average and Maximum (Percent Available) Percent used at AVG AVG 20% Memory AVG 80% [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 21

25 Percent used at MAX MAX 17% Memory MAX 83% [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 22

26 Disk Activity: Input/Output Operations (56 candidates) Disk IOPS (Read, Write, & Total) at Average 50,000 40,000 30,000 20,000 10,000 0 Disk IOPS (Read, Write, & Total) at Maximum 60,000 50,000 40, Reference Model - Unix/Linux: Disk IOPS AVG 47,231 42,541 47,196 46,927 1/7/2013 1/14/2013 1/21/2013 1/28/2013 IOPS AVG IOPS AVG AVG 2014 Reference Model - Unix/Linux: Disk IOPS MAX 50,944 44,378 49,359 49,706 Commentary The transition to centralized computing via virtualization places a significant dependency on centralized storage. Much of the functionality available from virtualization comes from its close links with the storage layer. The storage solution will need to meet the demands of the IOPs of the business. The Disk Operation consumption graphs show how the demands placed on the system vary from week to week. The high point of demand comes in week 4 when 50,944 IOPs were performed. This diminishes to its low point in week 2 when it is reduced to 44,378 IOPs. This represents a 6,566 IOPs variance from the low peak to the high peak. [Questionnaire Section 35 - Question 04] Additional Commentary - Storage IOPS. 30,000 20,000 10, /7/2013 1/14/2013 1/21/2013 1/28/2013 IOPS MAX IOPS MAX IOPS MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 23

27 Disk IOPS (Read) 2014 Reference Model - Unix/Linux: Disk IOPS Read 12,000 10,000 8,000 10,337 7,588 7,371 8,681 6,000 4,000 2, /7/2013 1/14/2013 1/21/2013 1/28/2013 IOPS MIN IOPS AVG IOPS 95th% IOPS MAX Disk IOPS (Write) 2014 Reference Model - Unix/Linux: Disk IOPS Write 45,000 40,000 35,000 30,000 25,000 20,000 15,000 10,000 5, ,391 37,223 41,995 42,047 1/7/2013 1/14/2013 1/21/2013 1/28/2013 IOPS MIN IOPS AVG IOPS 95th% IOPS MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 24

28 Disk IOPS (Total) Average and Maximum 2014 Reference Model - Unix/Linux: Disk IOPS Total [AVG + MAX] 60,000 50,000 40,000 30,000 20,000 10, ,944 47,231 44,378 49,359 42,541 47,196 1/7/2013 1/14/2013 1/21/ ,706 46,927 1/28/2013 AVG IOPS MAX Disk IOPS (Total) at Maximum 2014 Reference Model - Unix/Linux: Disk IOPS Total 60,000 50,000 50,944 44,378 49,359 49,706 40,000 30,000 20,000 10, /7/2013 1/14/2013 1/21/2013 1/28/2013 IOPS MIN AVG IOPS 95th% IOPS MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 25

29 Disk Activity: Volume (56 candidates) Disk Volume (Read, Write, & Total) at Average 2014 Reference Model - Unix/Linux: Disk VOL AVG (B/s) 977,093,788 1,000,000, ,404, ,041, ,574, ,000, ,000, ,000, ,000, /7/2013 1/14/2013 1/21/2013 1/28/2013 Disk Vol AVG Disk Vol AVG Disk Vol AVG Disk Volume (Read, Write, & Total) at Maximum 2014 Reference Model - Unix/Linux: Disk VOL MAX (B/s) 2,428,042,714 2,500,000,000 2,218,410,114 2,390,252,256 2,358,500,352 2,000,000,000 1,500,000,000 Commentary These graphs indicate the volume of data being read from or written out to the underlying storage. The centralized storage will need to be able to cope not only with the amount of transactions, but also the volume of data. The blend of the read and write transactions will have an effect on the actual storage design. Disk writes impose a much greater load on the storage subsystem and often is a limiting factor in commodity clouds. The Read and Write consumption graphs show how the demands placed on the system vary from week to week. The peak for Reads of 2,347,997,440 bytes per second is hit in week and the peak for Writes is hit in week at 543,021,044 bytes per second. [Questionnaire Section 35 - Question 05] Additional Commentary - Storage Data Volume. 1,000,000, ,000, /7/2013 1/14/2013 1/21/2013 1/28/2013 Disk Vol MAX Disk Vol MAX Disk Vol MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 26

30 Disk Volume (Read) 2014 Reference Model - Unix/Linux: Disk VOL Read (B/s) 2,500,000,000 2,332,766,720 2,154,120,826 2,347,997,440 2,280,903,168 2,000,000,000 1,500,000,000 1,000,000, ,000, /7/2013 1/14/2013 1/21/2013 1/28/2013 Disk Vol MIN Disk Vol AVG Disk Vol 95th% Disk Vol MAX Disk Volume (Write) 2014 Reference Model - Unix/Linux: Disk VOL Write (B/s) 600,000, ,000, ,000, ,000, ,185, ,832, ,021, ,563, ,000, ,000, /7/2013 1/14/2013 1/21/2013 1/28/2013 Disk Vol MIN Disk Vol AVG [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 27

31 Disk Volume (Total) Average and Maximum 2014 Reference Model - Unix/Linux: Disk VOL Total [AVG + MAX] 2,428,042,714 2,500,000,000 2,218,410,114 2,390,252,256 2,358,500,352 2,000,000,000 1,500,000, ,093,788 1,000,000, ,404, ,000, ,041, ,574, /7/2013 1/14/2013 1/21/2013 1/28/2013 Disk Vol AVG Disk Vol MAX Disk Volume (Total) at Maximum 2014 Reference Model - Unix/Linux: Disk VOL Total 2,500,000,000 2,428,042,714 2,218,410,114 2,390,252,256 2,358,500,352 2,000,000,000 1,500,000,000 1,000,000, ,000, /7/2013 1/14/2013 1/21/2013 1/28/2013 Disk Vol MIN Disk Vol AVG [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 28

32 Network Data Volume (55 candidates) Network Utilization (B/s) 2014 Reference Model - Unix/Linux: Network (B/s) 1,200,000,000 1,076,702,038 1,000,000, ,186, ,000, ,710, ,000, ,966, ,000, ,000, /7/2013 1/14/2013 1/21/2013 1/28/2013 NET MIN NET AVG NET 95th% NET MAX Network Utilization (Percent) 2014 Reference Model - Unix/Linux: Network (%) 6% 5% 5% 4% 4% 3% 3% 3% 2% 1% 0% 1/7/2013 1/14/2013 1/21/2013 1/28/2013 NET % MIN NET % AVG NET % 95th% NET % MAX Networking Commentary Most physical machines have a connection to the network via Gigabit network interface cards (NICs). When these machines are virtualized, they will share network interfaces via virtual networking. Therefore it is critical that the networking layer is not overloaded leading to contention for resources that could affect the perceived performance and availability of the virtual machines. Within the physical hosts used by virtualized infrastructures and the cloud, multiple physical networking interfaces (NICs) will be bonded or teamed to provide higher bandwidth trunks and redundancy to the core network infrastructure and this will then be allocated in a load balanced manner to virtual machines using algorithms defined within virtual network switches. The Reference Systems Inc. network was being utilized at its highest rate during week when 1,076,702,038 bytes per second was measured. This peak was reduced to 588,710,958 bytes per second during week. [Questionnaire Section 35 - Question 06] Additional Commentary - Network Data Volume. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 29

33 Network Utilization Average vs. Maximum (B/s) 2014 Reference Model - Unix/Linux: Network [AVG + MAX] (B/s) 1,200,000,000 1,000,000, ,000, ,000, ,000, ,000, ,076,702, ,186, ,710, ,966, ,967, ,747, ,642, ,488,109 1/7/2013 1/14/2013 1/21/2013 1/28/2013 NET AVG NET MAX Network Utilization Average vs. Maximum (Percent) 2014 Reference Model - Unix/Linux: Network [AVG + MAX] (%) 6% 5% 4% 3% 2% 1% 0% 0.9% 2.8% 0.9% 3.6% 0.9% 5.1% 0.9% 2.8% 1/7/2013 1/14/2013 1/21/2013 1/28/2013 NET % AVG NET % MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 30

34 Network Utilization at Maximum (B/s and % Available) 100% 80% 2014 Reference Model - Unix/Linux: MAX (B/s & %) 60% 40% 20,665,834,497 20,492,358,464 20,177,843,417 20,660,579,414 20% 0% 588,710, ,186,990 1,076,702, ,966,041 1/7/2013 1/14/2013 1/21/2013 1/28/2013 NET MAX NET MAX Network Utilization at Average and Maximum Percent used at AVG NET AVG 1% NET AVG 99% [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 31

35 Percent usedat MAX NET MAX NET MAX 4% 96% [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 32

36 Storage Capacities (56 candidates) Storage Utilization (GB) 25,000 20,000 15,000 10,000 5, Reference Model - Unix/Linux: Storage (GB) Storage Utilization (Percent) 24,295 24,299 24,304 24,304 1/7/2013 1/14/2013 1/21/2013 1/28/2013 MIN AVG 95th% MAX 80% 70% 60% 50% 40% 30% 20% 10% 0% 2014 Reference Model - Unix/Linux: Storage (% of Asset) 78% 78% 78% 78% Storage Commentary Virtualization increases the use of centralized storage in order to provide machine portability and high availability. Therefore it is necessary that enough storage is provisioned for the actual data as well as the projected growth and for advanced functionality like snapshots and replication facilities. Finally, it is critical to ensure that the necessary IOPs capacity is available to ensure required response times. Reference Systems Inc. had approximately 30.6 TB of storage allocated and of this 77.7% is utilized approaching 23.7 TB. The free space represents 22.3% of the available storage or 6.8 TB of storage. [Questionnaire Section 35 - Question 07] Additional Commentary - Storage Capacity. 1/7/2013 1/14/2013 1/21/2013 1/28/2013 Storage MIN Storage AVG Storage 95th% Storage MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 33

37 Storage Utilization Average vs. Maximum (GB) 2014 Reference Model - Unix/Linux: Storage (GB) 25,000 24,295 24,299 24,304 24,304 20,000 15,000 10,000 5, /7/2013 1/14/2013 1/21/2013 1/28/2013 MIN AVG 95th% MAX Storage Utilization Average vs. Maximum (Percent) 2014 Reference Model - Unix/Linux: Storage (% of Asset) 80% 70% 60% 50% 40% 30% 20% 10% 0% 78% 78% 78% 78% 1/7/2013 1/14/2013 1/21/2013 1/28/2013 Storage MIN Storage AVG Storage 95th% Storage MAX [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 34

38 Storage Utilization at Maximum (GB and Percent Available) 100% 80% 2014 Reference Model - Unix/Linux: MAX (B/s & %) 60% 40% 20,665,834,497 20,492,358,464 20,177,843,417 20,660,579,414 20% 0% 588,710, ,186,990 1,076,702, ,966,041 1/7/2013 1/14/2013 1/21/2013 1/28/2013 NET MAX NET MAX Storage Utilization at Average and Maximum Percent used at AVG AVG 22% Storage AVG 78% [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 35

39 Percent used at MAX MAX 22% Storage MAX 78% [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 36

40 7. Aggregate Resource Utilization over Time Aggregate CPU Utilization 400, , , , , , ,000 50, Reference Model - Unix/Linux: Aggregate CPU (MHz) Aggregate Memory Utilization 3,000,000 2,500,000 2,000,000 1,500,000 1,000, , Reference Model - Unix/Linux: Aggregate Memory (MB) [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 37

41 Aggregate Disk IOPS (Read) 12,000 10,000 8,000 6,000 4,000 2, Reference Model - Unix/Linux: Aggregate Disk IOPS (Read) Aggregate Disk IOPS (Write) 45,000 40,000 35,000 30,000 25,000 20,000 15,000 10,000 5, Reference Model - Unix/Linux: Aggregate Disk IOPS (Write) [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 38

42 Aggregate Disk Volume (Read) 2,500,000, Reference Model - Unix/Linux: Aggregate Disk Volume (Read) B/s 2,000,000,000 1,500,000,000 1,000,000, ,000,000 0 Aggregate Disk Volume (Write) 600,000, ,000, ,000, ,000, ,000, ,000, Reference Model - Unix/Linux: Aggregate Disk Volume (Write) B/s [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 39

43 Aggregate Disk IOPS (Total) 60,000 50,000 40,000 30,000 20,000 10, Reference Model - Unix/Linux: Aggregate Disk IOPS (Total) Aggregate Disk Volume (Total) 2,500,000, Reference Model - Unix/Linux: Aggregate Disk Volume (Total) B/s 2,000,000,000 1,500,000,000 1,000,000, ,000,000 0 [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 40

44 Aggregate Storage Utilization 25,000, Reference Model - Unix/Linux: Aggregate Disk Storage (GB) 20,000,000 15,000,000 10,000,000 5,000,000 0 Aggregate Network Utilization 1,200,000,000 1,000,000, ,000, ,000, ,000, ,000, Reference Model - Unix/Linux: Aggregate Network (B/s) [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 41

45 8. Week by Week Performance Analysis for Virtualization Overview The previous section concentrated on the combined loads of the differing workload areas. This section illustrates how this usage profile would affect the typical virtual machine profile and what the effect on a suggested host platform would be. Each area of critical importance to virtualization/consolidation is summarized to create the week s virtualization candidate profile, and this is modelled against the proposed host resources that form the virtualization platform. It shows the demand that the measured systems would place upon the consolidated design. The information is presented in a week by week basis to illustrate the performance trends. Resource Totals Per Server CPU Allocated MHz CPU Util. MHz Average Processor Speed Peak Av CPU MHz consumption VMware Peak Av CPU Utilization based on average % Memory Allocated MB Memory Used Memory Free Memory utilization % Disk I/0 Network (bytes/sec) Total CPU MHz available in all the candidates. The total literal MHz consumption of the entire estate based on each individual server s consumption. N/A The total MHz consumption of the servers based on literal calculation. The average MHz consumed based on the total MHz in the estate divided by the average %. This is not as accurate as the CPU utilized MHz (measured figure) The total allocated Mb of RAM in the estate The total memory used in the Estate The total amount of free memory not being utilized in the estate The percentage that the memory is utilized in the estate The number of disk Input/output operations per second in the entire estate. The total bytes per second transferred on the network interface cards of the entire estate. Total divided by the number of candidates The average of total literal MHz consumption of the entire estate based on each individual servers consumption The average speed of the processor installed in the servers Take the calculated MHz figure and then divide by the number of servers The average MHz consumed based on the total MHz in the estate divided by the average % divided by the number of candidate servers. This is not as accurate as the CPU utilized MHz (measured figure) The total allocated Mb of RAM in the estate divided by the number of servers The total memory used in the Estate divided by the number of servers The total amount of free memory not being utilized in the estate divided by the number of servers The percentage that the memory is utilized in the estate The number of disk Input/output operations per second in the entire estate divided by the number of servers The total bytes per second transferred on the network interface cards of the entire estate divided by the number of servers. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 42

46 Data Collection Periods Period One: Number of Candidates: 56 Period Two: Number of Candidates: 56 Period Three: Number of Candidates: 56 Period Four: Number of Candidates: 56 [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 43

47 Measured Performance Data Period 1 Resource (at MAX Utilization) Totals Per Server Number of Targets 56 Divided by # of Targets CPU: Allocated (MHz) 2,035,781 36,353 CPU: Utilization (MHz) ** 350,861 6,265 CPU: Utilization (Percent) 17.2% 17.2% CPU: Free at MAX (MHz) 1,684,920 30,088 Memory: Allocated (MB) 3,169,532 56,599 Memory: Utilization (MB) ** 2,643,511 47,206 Memory: Utilization (Percent) 83.4% 83.4% Memory: Free (MB) 526,021 9,393 Disk I/0PS (Read) 10, Disk I/0PS (Write) 43, Disk I/0PS (Total) ** 50, Disk Volume (Read) 2,332,766,720 41,656,549 Disk Volume (Write) 387,185,158 6,914,021 Disk Volume (Total) 2,428,042,714 43,357,906 Network: Allocated (Mbits) 170,036 3,036 Network: Utilization (B/s) ** 588,710,958 10,512,696 Network: Utilization (Percent) 2.8% 2.8% Network: Free 20,665,834, ,032,759 Storage: Allocated 31, Storage: Utilization (MB) 24, Storage: Utilization (Percent) 77.6% 77.6% Storage: Free 6, **Resource Unit Component [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 44

48 Measured Performance Data Period 2 Resource (at MAX Utilization) Totals Per Server Number of Targets 56 Divided by # of Targets CPU: Allocated (MHz) 2,035,781 36,353 CPU: Utilization (MHz) ** 321,046 5,733 CPU: Utilization (Percent) 15.8% 15.8% CPU: Free at MAX (MHz) 1,714,735 30,620 Memory: Allocated (MB) 3,169,532 56,599 Memory: Utilization (MB) ** 2,630,420 46,972 Memory: Utilization (Percent) 83.0% 83.0% Memory: Free (MB) 539,112 9,627 Disk I/0PS (Read) 7, Disk I/0PS (Write) 37, Disk I/0PS (Total) ** 44, Disk Volume (Read) 2,154,120,826 38,466,443 Disk Volume (Write) 454,832,486 8,122,009 Disk Volume (Total) 2,218,410,114 39,614,466 Network: Allocated (Mbits) 170,036 3,036 Network: Utilization (B/s) ** 762,186,990 13,610,482 Network: Utilization (Percent) 3.6% 3.6% Network: Free 20,492,358, ,934,973 Storage: Allocated 31, Storage: Utilization (MB) 24, Storage: Utilization (Percent) 77.7% 77.7% Storage: Free 6, **Resource Unit Component [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 45

49 Measured Performance Data Period 3 Resource (at MAX Utilization) Totals Per Server Number of Targets 56 Divided by # of Targets CPU: Allocated (MHz) 2,035,781 36,353 CPU: Utilization (MHz) ** 366,538 6,545 CPU: Utilization (Percent) 18.0% 18.0% CPU: Free at MAX (MHz) 1,669,243 29,808 Memory: Allocated (MB) 3,169,532 56,599 Memory: Utilization (MB) ** 2,645,193 47,236 Memory: Utilization (Percent) 83.5% 83.5% Memory: Free (MB) 524,339 9,363 Disk I/0PS (Read) 7, Disk I/0PS (Write) 41, Disk I/0PS (Total) ** 49, Disk Volume (Read) 2,347,997,440 41,928,526 Disk Volume (Write) 543,021,044 9,696,804 Disk Volume (Total) 2,390,252,256 42,683,076 Network: Allocated (Mbits) 170,036 3,036 Network: Utilization (B/s) ** 1,076,702,038 19,226,822 Network: Utilization (Percent) 5.1% 5.1% Network: Free 20,177,843, ,318,632 Storage: Allocated 31, Storage: Utilization (MB) 24, Storage: Utilization (Percent) 77.7% 77.7% Storage: Free 6, **Resource Unit Component [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 46

50 Measured Performance Data Period 4 Resource (at MAX Utilization) Totals Per Server Number of Targets 56 Divided by # of Targets CPU: Allocated (MHz) 2,035,781 36,353 CPU: Utilization (MHz) ** 301,543 5,385 CPU: Utilization (Percent) 14.8% 14.8% CPU: Free at MAX (MHz) 1,734,238 30,969 Memory: Allocated (MB) 3,169,532 56,599 Memory: Utilization (MB) ** 2,641,670 47,173 Memory: Utilization (Percent) 83.3% 83.3% Memory: Free (MB) 527,862 9,426 Disk I/0PS (Read) 8, Disk I/0PS (Write) 42, Disk I/0PS (Total) ** 49, Disk Volume (Read) 2,280,903,168 40,730,414 Disk Volume (Write) 343,563,776 6,135,067 Disk Volume (Total) 2,358,500,352 42,116,078 Network: Allocated (Mbits) 170,036 3,036 Network: Utilization (B/s) ** 593,966,041 10,606,536 Network: Utilization (Percent) 2.8% 2.8% Network: Free 20,660,579, ,938,918 Storage: Allocated 31, Storage: Utilization (MB) 24, Storage: Utilization (Percent) 77.7% 77.7% Storage: Free 6, **Resource Unit Component [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 47

51 9. Resource Utilization Analysis Overview This section provides an estimate of the number of MicroVMs (µvms) that would be required by Reference Systems Inc. µvms are a unit of measure that focus on the fundamental system resources of the environment. µvms measure Memory, CPU, IOPS, and Network in segments that are directly comparable to a virtualized system inventory. The µvm is designed to be technology and service provider independent: they show a measure of resource that is not favoring any specific vendor. The µvm also estimates the resource pool for the target environment: it avoids the allocation trap of estimating the number of virtual machines based on preconfigured products, presenting the optimal resource pool for a given environment without extraneous overhead for unused VM capacity. With this data, Reference Systems Inc. can estimate the potential requirement of any target virtualization technology or cloud service provider. Reference Systems Inc. can align the µvm resource pool with the product definition of the vendor, and quickly determine the appropriate allocation in the vendor s environment. At the end of each section, the highest consumer is indicated to illustrate the resource component that is the most consumed. Resource Totals Unit Value CPU Util. MHz The total MHz of CPU available in one µvm 200 MHz Memory Used The total MB of RAM available in one µvm 768 MB Disk I/0 Network (Mbit/s) The total number of Disk Input/Output Operations per Second available in one µvm The total Mbits per Second transferred on the NIC cards available in one µvm 40 IOPS 2 Mbit/s [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 48

52 µvm Consumption (56 candidates) The charts below illustrate how the resource consumption varies in relation to the µvm resource allocation Reference Model - Unix/Linux: µvm Requirement CPU Memory Disk I/O Network 0 1,755 1,606 1,833 1,508 2,685 2,672 2,687 2,683 1,110 1,274 1,234 1,243 2,355 3,049 2,376 1,000 2,000 3,000 4,000 4,307 5,000 1/7/2013 1/14/2013 1/21/2013 1/28/2013 Resource Unit (µvm) Component Max Consumption µvm Unit Value**** µvm Unit Total During Period CPU (Mhz) 366, MHz/µVM 1,833 Memory (MB) 2,645, MB/µVM 2,687 Disk I/OPS 50, IOPS/µVM 1,274 Network (B/s) 1,076,702,038 2 Mbps/µVM 4,307 µvm Commentary Over the measured period, the servers would require 4,307 µvms to meet the highest consuming component (Network). [Questionnaire Section 35 - Question 08] Additional Commentary - µvm Consumption. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 49

53 10. Potential Impact of Virtualization Proposed Virtualization Host Resources The previous section analyzed the resources currently used within the existing server estate. That information is utilized within this section to map this onto a consolidation design utilizing virtualization. The following table indicates the total available resource in each of the proposed VMware vsphere ESX hosts to run operate the virtual machines. The platform has been modelled with the following parameters: Most intensive processing week to provide CPU requirements Most intensive memory week to provide RAM requirements Additional host resource to cope with a failure to one host or for online maintenance Active/Active operation capability Provision to cater for 25% of estate utilizing Fault Tolerant facilities with the assumption of 2x resource consumption per server utilizing this facility Site A Virtualization Resources Main Data Center 33 Host(s) in Normal Operations Resource Host Total Capacity Site Capacity Host CPU Resource (1 CPU, 6 Core w/2600 MHz speed) (MHz) 11, ,656 Memory (GB) factored to 80% 205 6,758 Network (Gbps) Storage (Gbps) Site B Virtualization Resources Secondary Data Center 33 Host(s) in Normal Operations Resource Host Total Capacity Site Total Capacity Host CPU (1 CPU, 6 Core w/2600 MHz speed) (MHz) 11, ,656 Memory (GB) factored to 80% 205 6,758 Network (Gbps) Storage (Gbps) Combined Most Intensive Resource Utilization modelling Utilizing the performance data captured in the previous section, we take the combined load of each of the most intensive weeks to create the combined peak average load candidate. This is then modelled onto the proposed host resources to illustrate the likely utilization. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 50

54 Modelled Candidate with Combined Maximum Source Demands Typical Profile x 56 Candidates Resource Peak Modeled Candidate Maximum Utilization Processor Utilization (MHz) 6, ,538 Memory Utilization (MB) 47,236 2,645,193 Storage Allocation (GB) Storage Utilization (GB) Disk IOPS Read ,337 Disk IOPS Write ,391 Disk IOPS Total ,944 Disk Volume Read (B/s) 9,696, ,021,044 Disk Volume Write (B/s) 9,696, ,021,044 Disk Volume Total (B/s) 43,357,906 2,428,042,714 Network Bandwidth (B/s) 19,576,401 1,096,278,438 Storage Bandwidth (Disk Vol Rd + Wr) 19,393,609 1,086,042,088 Typical Profile of the Production Consolidation Platform Utilization (All Hosts Active/Active) Resource Maximum Utilization Site Total Capacity % Utilized #NAME? #NAME? #NAME? #NAME? ,656 0% #NAME? #NAME? #NAME? Memory (MB) 2,583 6,758 38% #NAME? #NAME? #NAME? Network (4 Gbps) 1,070, % #NAME? #NAME? #NAME? Storage (4 Gbps) 1,060, % #NAME? #NAME? #NAME? The table above shows the utilization of 33 host(s) at each site in an active / active configuration with standard operations and all hosts operating. Typical Profile of Production Consolidation Platform Utilization (All Hosts Active/Active Loss of One Host) Resource Maximum Utilization Site Total Capacity % Utilized #NAME? #NAME? #NAME? #NAME? 358 #NAME? #NAME? #NAME? #NAME? #NAME? Memory (GB RAM) 2,583 #NAME? #NAME? #NAME? #NAME? #NAME? Network (4 Gbps) 1,070,584 #NAME? #NAME? #NAME? #NAME? #NAME? [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 51

55 Resource Maximum Utilization Site Total Capacity % Utilized #NAME? #NAME? #NAME? Storage (4 Gbps) 1,060,588 #NAME? #NAME? #NAME? #NAME? #NAME? The table above shows the utilization of 32 host(s) at each site in an active/active configuration with standard operations and one host in maintenance mode (inactive). [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 52

56 11. Potential Impact of Cloud Infrastructure Proposed Cloud µvm Resources This section maps the resource configuration profiled in the previous section into the target cloud environment(s). The proposal has been modelled with the following parameters: Most intensive processing week to provide CPU requirements Most intensive memory week to provide RAM requirements Most intensive IOPs requirements Most intensive Networking Current Overall Resource Utilization Resource Total during Peak Period % Utilized µvms CPU (Mhz) 366, % 1,833 Memory (MB) 1,920, % 2,687 Disk I/OPS 1,826,416 N/A 1,274 Network (B/s) 1,767,340 N/A 4,307 [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 53

57 12. Environmental Impact Overview Virtualization has achieved widespread adoption in many organizations due to its ability to essentially do more with less and faster. Therefore virtualization is in essence a more efficient inherent strategy to pursue. One convenient side effect of pursuing virtualization is that it will undoubtedly require less utility than was in place previously. Until recently, hardware was notoriously inefficient as the priority was to maintain service levels, not necessarily to do so efficiently. Virtualization has highlighted the ability to eradicate huge areas of cost and inefficiency. Gaining much more notoriety today is the follow-on effect that if computing can be completed using less resource, then in turn with being more cost effective, it will also place less demand on the environment. Recent initiatives such as The Green Grid are extolling the virtue of designing the efficient data center and core to this is consolidation and virtualization. Existing Infrastructure Power Consumption Utility Demands Using a working average of a 500W power supply in each of the physical servers, the table below shows the current power demands and its related cooling requirements. The 500W is the stated maximum for the power supply and it has been factored down to 70% to allow for configuration variations within the server platform, this represents a conservative approach to assessing power consumption. Current Physical Platform for Initial Candidates All based on 24*7 Usage Resource Power (Watts per hour) Quantity Total KWPH Daily KW Consumption Yearly KWh #NAME? Servers 350 Reference Systems Inc. #VALUE! #VALUE! #VALUE! #VALUE! Air Con #VALUE! 1 #VALUE! #VALUE! #VALUE! #VALUE! Total #VALUE! #VALUE! #VALUE! #VALUE! The table above shows for 30 physical, (eliminating the estimated 70 virtual) candidates in scope. The power consumption total for this implementation is #VALUE! KW per year costing #VALUE! per year in utility costs. This is based on 1.2W of energy being consumed to provide the cooling for the 1 watt consumed. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 54

58 Impact of Virtualization Solution Power Consumption Power Consumption of the proposed Virtualization Live Platform All based on 24*7 usage). Resource Power (Watts per hour) Quantity Total KWPH Daily KW Consumption Yearly KWh #NAME? Servers ,178 $12, Storage 4, ,263 $9, Air Con 11, ,178 $12, Total ,619 $33, The above figures take into account the existing virtualized infrastructure; the number of hosts is reflective of what would be required to virtualize the 56 servers. These virtualization hosts are based on servers operating 835W power supplies to 100% resulting in the 835W power consumption per hour per server. In addition, centralized storage would be required, and a representative power consumption of a dual site implementation of a storage solution capable of meeting the requirements has been specified. To run the surveyed servers the 33 host(s) would utilize 101,178KW per year resulting in a likely cost of $12, This is then increased by the need for centralized storage which would require 77,263KW per year and cost $9, The cooling based on 1.2 Watts consumed to provide cooling for 1 watt would add an additional 101,178KW per year. This brings the total cost to $33, per year for electricity costs and cooling. Compared to the existing physical model this would save #VALUE! KW per year and #VALUE!. Carbon Comparisons The key to the reduction in carbon footprint emissions in relation to data centers is within the following areas: Improved efficiency of the power supplies to reduce the inefficiency when demand is low Reduction in the number of servers by utilizing the resources efficiently Reduction in the associated cooling requirements due to less emitted BTUs by less servers The carbon footprint impact of the candidates for consolidation is detailed in the comparison table below. Resource Total KWPY Carbon Calculation Carbon Footprint (Tons per Year) Footprint Physical #VALUE! #VALUE! #VALUE! Footprint Virtual 279,619 #DIV/0! #DIV/0! Reduction #VALUE! #VALUE! #VALUE! % Reduction #VALUE! #VALUE! #VALUE! [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 55

59 Calculation Key Typical gas fired power stations produce 440g of CO2 per KWPH, oil fired stations produce 650g of CO2 per KWPH coal fired stations 950g per KWPH. The above calculation uses 440g carbon dioxide per KWPH of electricity for the middle ground. The carbon footprint for Reference Systems Inc. would be reduced by #VALUE! per year when operating on a fully virtual platform. This does not take account for the additional reductions that would be possible by dynamic reduction in available host resources as demand decreases without any loss of service that would be possible once on a virtualized platform. This is a feature that is now fully supported, and host servers can be powered on and off to meet varying demand. The figure being currently estimated is an additional 20% reduction in consumption related to the virtualization host systems. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 56

60 Impact of Cloud Solution Power Consumption Power Consumption of the proposed Cloud Platform All based on 24*7 usage). Resource Power (Watts per hour) Quantity Total KWPH Daily KW Consumption Yearly KWh #NAME? Servers $0.00 Storage 4, ,263 $9, Air Con $0.00 Total ,263 $9, The above figures take into account the existing virtualized infrastructure; the number of hosts is reflective of what would be required to virtualize the 56 servers on the cloud platform. These virtualization hosts are based on servers operating 835W power supplies to 100% resulting in the 835W power consumption per hour per server. In addition, centralized storage would be required and a representative power consumption of a dual site implementation of a storage solution capable of meeting the requirements has been specified. Carbon Footprint Comparisons The key to the reduction in carbon footprint emissions in relation to data centers is within the following areas: Improved efficiency of the power supplies to reduce the inefficiency when demand is low Reduction in the number of servers by utilizing the resources efficiently Reduction in the associated cooling requirements due to less emitted BTUs by less servers The carbon footprint impact of the candidates for consolidation is detailed below. Resource Total KWPY Carbon Calculation Carbon Footprint (Tons per Year) Footprint Physical #VALUE! #VALUE! #VALUE! Footprint Cloud 77,263 #DIV/0! #DIV/0! Reduction #VALUE! #VALUE! #VALUE! % Reduction #VALUE! #VALUE! #VALUE! Calculation Key Typical gas fired power stations produce 440g of CO2 per KWPH, oil fired stations produce 650g of CO2 per KWPH coal fired stations 950g per KWPH. The above calculation uses #DIV/0!g carbon dioxide per KWPH of electricity for the middle ground. The carbon footprint for Reference Systems Inc. would be reduced by #VALUE! per year when operating on a cloud services platform. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 57

61 13. Report Signoff Report Signoff Project: Reference Project I agree with and accept the contents of the report version 1.0, dated Thursday, January 23, Reference Systems Inc. Representative Job Title: Signed: Date: For and on behalf of Reference Systems Inc.: Print Name: Virtustream Project Lead Print name: Signed: Date: [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 58

62 14. Target List Target Name Operating System OS Version CPU Speed (MHz) Total CPU (MHz) Disk Capacity (MB) Memory (MB) Network (MBits) Target 004 CentOS release 5.8 (Final) el5 2,394 19, ,971 3,948 1,000 Target 005 CentOS release 5.8 (Final) el5 2,394 38, ,085 7,982 2,000 Target 006 CentOS release 5.7 (Final) el5 2,394 9,576 38,704 2,010 1,000 Target 007 CentOS release 5.8 (Final) el5 2,400 38,402 93,731 7,972 6,000 Target 008 CentOS release 5.8 (Final) el5 2,400 38,402 1,404,026 7,972 6,000 Target 009 CentOS release 4.5 (Final) ELsmp 1,867 7,467 72,671 1,010 1,000 Target 010 CentOS release 4.5 (Final) ELsmp 1,867 7,467 72,671 1,010 5,000 Target 011 CentOS release 4.5 (Final) ELsmp 1,596 6, ,377 4,051 1,000 Target 026 CentOS release 5.7 (Final) el5 2,494 79, ,794 16,049 2,000 Target 027 CentOS release 5.8 (Final) el5 2,500 5,000 63,528 12,010 1,000 Target 028 CentOS release 6.3 (Final) el6.x86_64 2,500 2,500 49, ,000 Target 029 CentOS release 5.8 (Final) el5 2,500 5,000 63,528 12,010 1,000 Target 030 CentOS release 5.6 (Final) el5 2,833 90, ,084 16,052 2,000 Target 031 CentOS release 5.6 (Final) el5 2,328 74, ,585 12,009 2,000 Target 032 CentOS release 5.6 (Final) el5 2,328 74, ,571 7,981 2,000 Target 033 CentOS release 5.6 (Final) el5 2,494 79, ,678 12,009 2,000 Target 034 CentOS release 6.2 (Final) el6.x86_64 2,833 5,667 97,815 2,006 10,000 Target 035 CentOS release 5.5 (Final) el5PAE 2,993 11, ,212 7,986 10,000 Target 036 CentOS release 5.5 (Final) el5 2,826 2, ,830 3,949 10,000 [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 59

63 Target Name Operating System OS Version CPU Speed (MHz) Total CPU (MHz) Disk Capacity (MB) Memory (MB) Network (MBits) Target 037 CentOS release 5.6 (Final) el5 2,494 9,975 61,776 3,568 2,000 Target 038 CentOS release 5.6 (Final) el5 2,826 2,826 61,436 2,889 2,000 Target 039 CentOS release 5.4 (Final) el5PAE 2,826 11, ,215 12,050 1,000 Target 040 CentOS release 5.8 (Final) el5PAE 2,826 45, ,335 7,986 1,000 Target 045 Target 046 Target 047 Target 048 Target 049 Target 050 Target 051 Target 052 Red Hat Enterprise Linux Server release 5.7 (Tikanga) Red Hat Enterprise Linux Server release 5.7 (Tikanga) Red Hat Enterprise Linux Server release 5.7 (Tikanga) Red Hat Enterprise Linux Server release 5.7 (Tikanga) Red Hat Enterprise Linux Server release 5.7 (Tikanga) Red Hat Enterprise Linux Server release 5.7 (Tikanga) Red Hat Enterprise Linux Server release 5.7 (Tikanga) Red Hat Enterprise Linux Server release 5.7 (Tikanga) el5 2,533 40, ,958 64,326 1, el5 2,533 40, ,959 64,327 1, el5 2,533 40, ,958 64,326 1, el5 2,534 40, ,958 64,326 1, el5 2,533 40, ,958 64, el5 2,533 40, ,958 64,325 3, el5 2,533 40, ,958 64,326 1, el5 2,533 40, ,958 64,325 3,000 [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 60

64 Target Name Operating System OS Version Target 069 CentOS Linux release 6.0 (Final) el6.x86_64 CPU Speed (MHz) Total CPU (MHz) Disk Capacity (MB) Memory (MB) Network (MBits) 3,167 3,167 6, ,000 Target 070 CentOS release 5.4 (Final) el5 3,167 3,167 16,409 1,002 1,000 Target 071 CentOS release 6.2 (Final) el6.x86_64 2,394 4,788 99, ,000 Target 072 CentOS release 4.5 (Final) ELsmp 1,596 6, ,427 4,051 4,000 Target 073 CentOS release 4.6 (Final) ELsmp 2,394 2,394 31, ,000 Target 081 Target 082 Target 083 Target 084 Target 085 Target 086 Target 087 Red Hat Enterprise Linux AS release 4 (Nahant Update 6) Red Hat Enterprise Linux AS release 4 (Nahant Update 6) Red Hat Enterprise Linux AS release 4 (Nahant Update 6) Red Hat Enterprise Linux AS release 4 (Nahant Update 6) Red Hat Enterprise Linux Server release 5.5 (Tikanga) Red Hat Enterprise Linux Server release 5.5 (Tikanga) Red Hat Enterprise Linux Server release 5.5 (Tikanga) ELlargesmp 2,660 63,837 2,567,940 96,672 1, ELsmp 2,926 23,407 1,400, ,692 1, ELsmp 2,926 23,407 1,400, ,692 3, ELlargesmp 2,394 57,454 1,061,040 64,353 1, el5 2,660 31, ,632 64,448 5, el5 2,660 31, ,500 64,448 5, el5 2,660 31, ,500 64,448 5,000 [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 61

65 Target Name Operating System OS Version Target 088 Red Hat Enterprise Linux Server release 5.5 (Tikanga) CPU Speed (MHz) Total CPU (MHz) Disk Capacity (MB) Memory (MB) Network (MBits) el5 2,660 31, ,500 64,448 2,000 Target 089 Red Hat Enterprise Linux Server release 5.5 (Tikanga) el5 2,793 67, ,392 32,179 2,000 Target 090 Red Hat Enterprise Linux Server release 6.1 (Santiago) el6.x86_64 1,200 38, , ,141 2,000 Target 091 Red Hat Enterprise Linux Server release 6.1 (Santiago) el6.x86_64 1,200 38, , ,141 2,000 Target 092 Red Hat Enterprise Linux Server release 6.1 (Santiago) el6.x86_64 1,200 38, , ,141 2,000 Target 093 Red Hat Enterprise Linux Server release 6.1 (Santiago) el6uek.x86_64 2, ,174 3,267, ,455 6,000 Target 094 Red Hat Enterprise Linux Server release 6.1 (Santiago) el6uek.x86_64 2, ,213 3,267, ,455 14,000 Target 095 Red Hat Enterprise Linux Server release 6.1 (Santiago) el6uek.x86_64 2, ,182 3,267, ,455 5,000 Target 096 Target 097 Red Hat Enterprise Linux Server release 5.5 (Tikanga) Red Hat Enterprise Linux Server release 5.5 (Tikanga) el5 1,596 19, ,932 96,669 5, el5 1,596 19, ,932 96,669 5,000 [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 62

66 Target Name Operating System OS Version Target 098 Red Hat Enterprise Linux Server release 5.5 (Tikanga) CPU Speed (MHz) Total CPU (MHz) Disk Capacity (MB) Memory (MB) Network (MBits) el5 1,596 19, ,932 96,669 5,000 Target 099 Target 100 Red Hat Enterprise Linux Server release 5.3 (Tikanga) Red Hat Enterprise Linux Server release 5.3 (Tikanga) el5xen 2,527 40, ,392 31,729 3, el5xen 2,527 40, ,392 31,728 2,000 [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 63

67 15. About Virtustream, Inc. Virtustream, Inc. is a global infrastructure services firm committed to providing our clients pioneering and impartial thought leadership in defining Information Technology strategies. Our solutions enable clients to actualize the enterprise cloud by providing strategy, integration and managed services leveraging virtualization technologies, and our proprietary, secure platform, xstream. Virtustream delivers efficient infrastructure solutions backed by guaranteed service levels and industry leading resourcebased pricing models built upon the company s pillars of service excellence including Professional Services, Cloud Services and Managed Service solutions. Virtustream has been providing specialist independent IT consultancy in all areas of virtualization from the data center to the desktop for the last 10 years. We have won numerous industry awards in recognition of the detail and quality of our work including being awarded VMware s EMEA Consultancy Partner of the Year and the ICT - Green IT initiative of the year award. Our customers range from the smaller business to the larger enterprises, charities and government organizations including Domino Foods, Johns Hopkins, National Education Association, The World Bank, Federal Reserve IT, National Institutes of Health, Allied Irish Bank, EMC and VMware. The XStream Platform The xstream platform is unique to Virtustream and has been built and developed in-house using internal IP accumulated over 9 years of client engagements and including bespoke automation scripts to create a highly scalable, efficient secure and resilient infrastructure to support our clients. Virtustream has developed the platform to be vendor agnostic to leverage best of breed hardware and software and is made available as a dedicated or shared platform, either managed or unmanaged, in our data centers or on a client s premises. This provides our clients with the flexibility of options, so that we can host critical applications on any platform with guaranteed committed level of resources defined by units. These computing units are calculated through CPU, RAM, IOPs, Network, and requirements for High Availability and Disaster Recovery. The whole environment is also tuned to optimize ERP applications such as SAP, Oracle and Microsoft and is easily accessed via a personalized Silverlight portal. Solution Overview The xstream platform features an N+2 cluster design for high availability within the data center. Similarly, the platform features resilience across two data centers. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 64

68 16. Appendix - Virtualization Explained Overview The viability of pursuing a virtualization strategy is being investigated as a solution to the objectives identified in the previous section. This report discusses the issues faced and how server virtualization could offer a solution to the issue. Various Forms of Virtualization Infrastructure Area Description Main Benefits Server The most adopted to date area of virtualization where a software layer, hypervisor, is installed onto industry standard hardware enabling it to host partitioned virtualized containers or virtual machines. The underlying resource in the host is shared out to the virtual machines, and this has been successful in combating Moore s law as hardware advances outstrip the ability to utilize them effectively. Massive reductions in capital expenditure on hardware Energy consumption reductions Increased availability Improved responsiveness to the organization Centralized management of resources Desktop Designed to repeat the success of server virtualization, Virtual Desktop Infrastructure (VDI), is directed at repeating the success of Terminal Server-based desktop solutions (Microsoft Terminal Services and Citrix Presentation Server) but with improved compatibility and increased management. Again host servers are utilized to provide virtual machines running desktop operating systems, and users connect to these via Connection Brokers which decide what image is presented to which user and what functionality is provided. Centralized Management of the Desktop Security Anywhere computing Rapid deployment of applications or upgrades Service continuity of desktop inherited from underlying infrastructure Application Application management is one of the biggest challenges to an I.T. department and has some of the greatest impact on an organization. Virtualized applications are presented to the end user in run time containers that are decoupled from the underlying operating system. Therefore applications can be streamed to users upon request, application upgrades can be performed on the source application at the data center, and this is automatically deployed at next request. Build consistency of the desktop is maintained as applications are not installed Complementary technology to VDI enabling improved reliability of the application stack. Removal of application conflicts leading to reduced application roll out times [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 65

69 Infrastructure Area Description Main Benefits Storage The foundation on which all virtualization depends is the storage layer, and Storage Virtualization enables the data residing within the physical storage devices to be accessed via a virtualization layer which enables an independence from the underlying hardware. Data is able to be replicated to multiple locations, and applications access the data via the virtualized instance, meaning data can traverse physical boundaries. It also enables greater efficiency in the consumption of storage as virtual volumes can be thin provisioned, meaning they are built for future requirements but only consume what is required for now. Data availability as dependency on the underlying physical resource is reduced massively. Performance can be increased as physical disks can all participate in the I/O demands Just in time provisioning through thinly provisioned data volumes Reduced power consumption by reduced footprint that is only increased as real data is required Network Covering both the traditional network-based layer used for traffic between devices and the core infrastructure communication layers, such as storage area networks. Network virtualization again takes the concept of consolidation and abstraction so that fewer physical interface cards are required and instead software layers control the traffic flow and overlay Quality of Service (QOS) to enable network interfaces to be shared. Technology area is fast becoming known as Convergence. Simplified connection networks Guaranteed service levels Share available resources between different networks Improved manageability What is Virtualization? Virtualization is not a new technology as it has been utilized within mainframe and mid-range machines since its invention by IBM over 30 years ago. It is relatively new in the area of x86 server-based technologies. It has been established in production use however for over 5 years. In the last two years, adoption rates have soared with most organizations attempting to exploit its benefit. Virtualization places a thin layer of software onto hardware to allow the physical resource to be dynamically segmented and presented to multiple server instances that run within the virtualization software. It effectively creates sub-systems dedicated to services much in the same way as mid-range and mainframe solutions and enables physical resources to be shared to multiple servers. This design aspect has been adopted by other areas of I.T. infrastructure and is now incorporated into the storage layer, application layer and workstation layer. Why is it being adopted by so many Inefficiency within the x86 physical models has been the primary driver. The reason that the issue exists in the first place is the typical deployment model for x86-based servers dictates that one service is deployed onto one server chassis, resulting in improved reliability (due mainly to the fact x86 operating systems have not traditionally multi-tasked effectively) but increased inefficiency. Hardware advances have exacerbated the issue as this has widened the gap between the resource available and the ability to utilize it; this has led to typical averages of less than 10% utilization with regard to processor utilization. This has now been expanded where the benefits witnessed in the server sector are now being examined in the application, storage and desktop models. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 66

70 Is it just another of the latest trends from IT that will be out of fashion in a year? IT often suffers from appearing to follow the latest developments only to repeat the process within a year when the latest new technology surfaces, leading to wastage in expenditure and a lack of cohesive strategy. When virtualization on x86 first appeared approximately six years ago it created significant interest but also attracted much skepticism. The landscape for virtualization today however presents a picture of huge adoption rates and acceptance as an established mainstream deployment solution. The main driving force behind this growth has been the delivery on the benefits promised by early adopters leading to exceptionally high levels of satisfaction by those organizations who have made the migration. The success of the technology has been based not on the technical benefits it offers IT services divisions but on the fundamental changes it has made to the operational effectiveness of those who have adopted it. Hardware manufacturers have changed their solutions to take advantage of the benefits of virtualization and as this trend progresses many industry observers comment that it will reach a point where virtualization is built into the hardware platform itself and will be a power-on option. Headline Benefits of Server Virtualization Issue Virtualization Solution Benefit Data Center Space Data Center Costs Operational efficiencies Service Levels Environmental Benefits Planning for future requirements Allow more services to operate on a physical platform by sharing underutilized resources to multiple servers By using the physical resources more efficiently, the asset is worked to higher levels (typically taking utilization from sub 10% towards 80%); and therefore, less hardware is required which reduces the associated costs. Servers that were previously hardware-based are presented as software; and therefore, deployment of servers can be achieved far more rapidly and reliably. Typical commissioning of a server can take 2 days currently (operating system and patching only), and this can be reduced to minutes by deploying standardized templates of virtual servers that are hardware independent. Aligning business service levels to hardware-based solutions can be complex and costly. Areas such as business continuity and high availability can result in duplication of resources. Virtualization has such solutions built into the underlying architecture; and therefore, solutions can be standardized and costs reduced. High power and cooling consumption of underutilized servers results in higher carbon emissions. Virtualization can significantly reduce these. With the established inefficiencies of the physical model, this places a higher emphasis on accurate predictive planning. While Virtualization does not advocate less diligence in the planning process, it does enable greater flexibility by being able to share resources and prioritize allocation based on service level decisions. Less hardware working more efficiently. Less data center space required. Power, cooling, space and maintenance costs can be significantly reduced. Server deployment can be completed in significantly less time and with higher degrees of standardization. Costly high availability solutions can be reduced in expenditure and complexity and made available to entire data centers. Massive savings in carbon emissions Far more flexible and dynamic solution more in line with the nature of organizations on demand model [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 67

71 Issue Virtualization Solution Benefit Dynamic Provisioning Virtualization through all its forms allows the dynamic management of physical resources to meet the demand; this allows organizations to build for now but expand effortlessly for the future. Matching I.T. strategy to organizational agility. What Should or Should Not Be Virtualized Virtualized Physical Virtualization Solution Any server where the measured performance can be combined with other servers to not cause a combined issue on the consolidation platform or where licensing proves to be prohibitive to do so. It is recommended to maintain at least one instance of the following type of servers: Active Directory Server DNS server Backup server Virtual Center Server Older versions of Citrix servers can present performance issues if running on Windows 2000 Prohibitive licensing models based on host resources not virtual server Reason Servers should always be measured for their suitability if transferring from existing physical platform as this can show underlying existing issues in the physical configuration that would mask where the issue resides if problems are encountered. By maintaining these outside the consolidation platform there will always be one instance that can be brought up in isolation or first during a sequence of power up to enable critical facilities for other services such as authentication or DNS. Virtual Center is a recommendation of Virtustream as it does not see the wisdom in having the management platform for virtualization to be held within the services it is managing. Vendor Criteria It is important that when choosing a platform vendor that the features and cost are evaluated in relation the stated goals. By evaluating the vendors against each other, it is possible to make the correct decision of which platform will provide the long term technology to deliver the TCO and ROI figures. When comparing the vendors, it is important to look at the near future to what will be available as this can influence any tactical decisions in the short term to move to an intermediate version of the vendor to allow migration to the preferred platform. This design is based on the current or release imminent from the vendors in the next 2-3 months as this is the realistic start of an implementation timeline. The vendors for comparison in this assessment were Microsoft Hyper-V R2 and VMware vsphere. This section looks at the overall feature comparison and additional products that are required to deliver the overall strategic design and then outlines an indicative costing for comparison. [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 68

72 Base Feature Comparison Microsoft Hyper-V R2 VMware vsphere Enterprise Plus CPU Capacity (Host) 32 cores (improved) 64 cores (improved) CPU Capacity (VM) 4 8 (improved) Memory Capacity (VM) 64GB 255GB (improved) Hot Add Resources (VM) Disk (new) CPU, Disk, Memory (improved) Hot Migration (VM) Yes (new) Yes Fault Tolerance (VM) No Yes (new) Storage migration Requires storage vendor support Yes (improved) Host baselines MS Server Management Suite Enterprise Yes (new) Patching / compliance MS Server Management Suite Enterprise Yes (improved) Non Windows OS support Yes* Yes** Automation platform Vizioncorev Control(new product) Yes (improved) Number of VMs 256 or resource limit (improved) 256 or Resource limit (improved) Thin provisioning Requires storage vendor support Yes (new) Dynamic Resourcing No Yes Dynamic Power No Yes (improved) HA Yes Yes Shared File System Yes (new) Yes Enterprise Management MS Server Management Suite Enterprise Yes P2V MS Server Management Suite Enterprise Yes Dynamic I/O Yes (new) No * Microsoft Windows CPU, 2003x86 1 or CPU, 2003 x64 1 or 2 CPU, 2008 x86 1, 2 or 4 CPU, 2008 x64 1, 2 or 4 CPU, Vista x86 1 or 2 CPU, Vista x86 1 or 2 CPU, Vista x64 1 or 2 CPU, XP Pro x86, XP Pro x64. Linux SUSE Enterprise Server 10 1 CPU ** Microsoft Windows PXE, 2008, Vista, 2003, XP, 2000, NT4, CentOS 5, Red Hat Enterprise 2.1-5, Red Hat, SUSE Enterprise Desktop 10-11, SUSE Enterprise Server 8-11, UBUNTU , Netware , Solaris 10 [Reference Systems Technology Incorporated] Virtustream Advisor Assessment Report 69

73 17. Appendix - Cloud Explained Overview Cloud computing is conceptually the provision of computing resources in an abstracted form via a delivery model that enables dynamically scalable resources provided as a service to an organization. Cloud solutions feature a significantly higher percentage of virtualized resources than those seen in most corporately provided infrastructures. The Cloud market is fast developing and the term Cloud is being used in an increasing number of diverse ways, often causing confusion in an emerging sector leading to a lack of clarity in what it is actually being offered and indeed what it actually delivers. Breaking through the hyperbole, a more accurate technical definition is a computing capability that is presented through an abstraction layer between the computing resource and the actual underlying technical architecture. This enables an on demand service connected to a shared pool of configurable resources that can be rapidly provisioned with minimal management overhead. Cloud services usually comprise of five essential characteristics, on demand self-service, broad network access, resource pools, rapid elasticity and a measured service. In most cases, organizations do not own the underlying resources; instead a model-based on rental usage is deployed paying only for resources that they use. Many cloud-computing offerings employ the utility computing model, which is analogous to how traditional utility services (such as electricity) are consumed, whereas others bill on a subscription basis. Different Cloud Solutions Cloud Category Public Cloud Hybrid Cloud Private Cloud Description Public cloud is what is typically envisaged when viewed in a mainstream sense. Resources are available easily, and to all, via the internet with entry to the solution sometimes as simple as a credit card transaction. The Public Cloud is typically self-service from a provisioning sense with clients presented a set of standardized infrastructure offering. Billing is typically based on a consumption utility model where up time, disk resource and performance metrics are factored in. The Hybrid Cloud is designed to combine the benefits of a series of solutions possibly from multiple providers; some view it as the logical solution for many enterprises where the requirements are too varied and complex to logically fit in one sector. An example could be that for disaster recovery services; an organization utilizes externally presented cloud resources for use in the case of a disaster invocation. The Private and Internal clouds are in actuality highly optimized environments that organizations operate internally often mirroring external cloud concepts. The concept behind them is that they offer the benefits offered by an external cloud while possibly avoiding some of the perceived pitfalls. The negatives with the approach are organizations are still left with the buy, build, manage and maintain models that can make other cloud solutions look so appealing. 70 Virtustream Advisor Assessment Report [Reference Systems Technology Incorporated]

74 Reference Project Cloud Services Service Provider Options In the above illustration it is easy to see the evolutionary scale of control as an organization moves through the service level offering from cloud services. In the traditional dedicated IT model on the first column we see the service level extending to the Network element of the technology stack, where most organizations rely on a service provider of some variety to provide their network links to locations. Progressing to the next column, that of the traditional Hosting Provider, we find it shifting in the areas of storage and server where the responsibility is shared. Above the server layer the client is still responsible for the virtual machines and everything that operates on them. Moving into Infrastructure as a Service (IaaS) the client has handed over responsibility solely to the provider in the areas of Network, Storage and Server and has joint responsibility for the virtual machine layer, everything else is controlled by the client. Platform as a Service (PaaS) extends this responsibility further still with the service provider sharing responsibility for the application layer and the actual Virtual Machines becoming part of a wider services agreement. In the final iteration, Software as a Service (SaaS) the client has delegated responsibility for the entire stack with them maintaining only shared control over the critical data layer. 71 Virtustream Advisor Assessment Report [Reference Systems Technology Incorporated]