value proposition for ibm power systems comparing costs of ibm powervm and x86 vmware for enterprise...

April 2009

MANAGEMENT BRIEF

VALUE PROPOSITION FOR IBM POWER SYSTEMS

Virtualization Impact for Enterprise UNIX and Linux Server Infrastructures

International Technology Group 4546 El Camino Real, Suite 230

Los Altos, California 94022-1069 Telephone: (650) 949-8410 Facsimile: (650) 949-8415 Email: [email protected]

ITG

Copyright © 2009 by the International Technology Group. All rights reserved. Material, in whole or part, contained in this document may not be reproduced or distributed by any means or in any form, including original, without the prior written permission of the International Technology Group (ITG). Information has been obtained from sources assumed to be reliable and reflects conclusions at the time. This document was developed with International Business Machines Corporation (IBM) funding. Although the document may utilize publicly available material from various sources, including IBM, it does not necessarily reflect the positions of such sources on the issues addressed in this document. Material contained and conclusions presented in this document are subject to change without notice. All warranties as to the accuracy, completeness or adequacy of such material are disclaimed. There shall be no liability for errors, omissions or inadequacies in the material contained in this document or for interpretations thereof. Trademarks included in this document are the property of their respective owners.

International Technology Group i

TABLE OF CONTENTS

EXECUTIVE SUMMARY 1

COST PICTURE 2 UNIX Server Infrastructures 2

IT Costs 2 Linux Server Infrastructures 5

IT Costs 5 Costs of Downtime 5

New UNIX System Deployments 6 Cost Comparisons 6 Costs of Downtime 9

DETAILED DATA 10 UNIX Server Infrastructure Comparisons 10

Installations and Scenarios 10 IT Costs 11 Costs of Downtime 13

Linux Server Infrastructures 13 Installations and Scenarios 13 IT Costs 14 Costs of Downtime 15

New UNIX System Deployments 15 Installations and Scenarios 15 IT Costs 17 Costs of Downtime 18

Cost Assumptions 19 Server Costs 19 Personnel Costs 19

List of Figures 1. Conventional and Power Virtualized UNIX Server Scenarios: Five-year Operating Costs 2 2. Conventional and Power Virtualized UNIX Server Scenarios: Five-year Overall Costs

(Virtualized Scenarios include Operating and Acquisition Costs) 3 3. Five-Year I/O-related Costs for Partitioned IBM Power Servers:

Averages for All Power Virtualized Scenarios 4 4. Conventional and Power Virtualized UNIX Server Scenarios: Five-year Costs of Downtime 4 5. x86 and Power Linux Server Scenarios: Average Five-year Costs for All Installations 5 6. x86 and Power Linux Server Scenarios: Five-Year Costs of Downtime 6 7. New UNIX System Deployments: Five-year IT Costs 7 8. Numbers of Hard Partitions Supported: HP and Sun UNIX Server Platforms 8 9. New UNIX System Deployments: Five-year Costs of Downtime 9 10. Profile Installations and Scenarios for UNIX Server Infrastructure Cost Comparisons 10 11. Cost Breakdowns for UNIX Server Infrastructure Comparisons 12 12. Profile Installations and Scenarios for Linux Server Infrastructure Cost Comparisons 13 13. Cost Breakdowns for Linux Server Infrastructure Comparisons 14 14. Profile Installations and Scenarios for New UNIX System Deployments Comparisons 15 15. Cost Breakdowns for New UNIX System Deployment Comparisons 18

International Technology Group 1

EXECUTIVE SUMMARY Server virtualization has emerged of one the critical technologies of our time. The potential clearly exists to significantly reduce costs and complexities of enterprise server infrastructures, to create new flexibility in server reconfiguration and provisioning, and to improve availability and other variables of service quality for server-based applications throughout organizations.

Although industry debate tends to focus on the application of VMware and equivalents to x86 server consolidation, in reality UNIX server virtualization technologies are – by wide margins – more sophisticated and mature than their x86 counterparts.

The Power platform is, in turn, the recognized industry leader in UNIX server virtualization. Because the technologies it employs have evolved rapidly over the last few years, even many longstanding Power users have yet to fully exploit its potential.

This report deals with this potential. Specifically, it looks at three opportunities to reduce IT costs through use of latest-generation IBM Power servers and PowerVM technology:

1. By replacing existing UNIX server infrastructures with virtualized Power servers, five-year operating costs may be reduced by an average of 61 percent.

Comparisons are between the use of diverse multivendor bases of UNIX servers, which are representative of most large organizations today, and current-generation Power servers exploiting the full potential of PowerVM virtualization capabilities.

If costs of acquiring new Power server hardware and systems software to realize these savings are allowed for, overall costs are still significantly lower. Five-year costs, including acquisition and operating costs, average 54 percent less.

2. By replacing existing Linux server infrastructures with virtualized Power servers, five-year operating costs may be reduced by an average of 58 percent.

Comparisons are between Linux applications deployed on distributed x86 servers and on current-generation Power servers again exploiting the potential of PowerVM capabilities. If acquisition costs are included in Power virtualized scenarios, five-year costs average 52 percent less.

3. For large-scale new UNIX system deployments, five-year acquisition and operating costs for use of Power servers average 43 percent less than those for Hewlett-Packard Integrity and 60 percent less than those for Sun Microsystems M and T series servers.

Comparisons are between database-intensive systems with high levels of virtualization deployed on Power servers using PowerVM; on HP Integrity servers using the company’s Virtual Server Environment (VSE); and on Sun servers employing dynamic domains, Logical Domains (LDoms), and Solaris Containers and Zones technologies.

Use of Power servers and PowerVM may also result in lower costs of downtime – meaning costs incurred by organizations due to disruptions affecting business processes and user productivity – than for existing UNIX and Linux server infrastructures, and for new UNIX system deployments. These results are outlined in the following section.

Additional information on the methodologies and assumptions employed, along with breakdowns of installations, configurations and costs are presented in the Detailed Data section of this report.


COST PICTURE

UNIX Server Infrastructures

IT Costs

This set of five-year cost comparisons is based on three composite profile installations in large financial services, manufacturing and retail companies with between $15 billion and $35 billion in revenues. For each profile installation, two sets of scenarios were developed:

1. Conventional scenarios. These include diverse bases of between 78 and 242 UNIX servers, including a variety of HP, IBM, Silicon Graphics and Sun models. The demographics of these server bases and the efficiency with which they are used are representative of most large organizations today. Virtualization is employed to a limited degree.

2. Power virtualized scenarios. In these, the same applications and workloads run on current-generation Power servers exploiting the full potential of PowerVM virtualization capabilities.

Five-year operating costs for Power virtualized scenarios range from 54 to 65 percent less, and average 61 percent less than those for conventional equivalents. Figure 1 summarizes these results.

Figure 1 Conventional and Power Virtualized UNIX Server Scenarios: Five-year Operating Costs

Operating costs, in this context, include hardware maintenance, update subscriptions and support for systems and database software, personnel for system administration-related functions, and facilities costs for data center occupancy, power and cooling.

Database costs are for Oracle databases and management tools. Calculations reflect the change in Oracle database pricing for POWER6-based servers in March 2009.


Lower costs were enabled through multiple virtualization capabilities. Both forms of partitioning supported by PowerVM – firmware-based logical partitions (LPARs) and software-based Workload Partitions (WPARs) – were employed aggressively to achieve high levels of server consolidation for diverse applications, databases and workloads.

Consolidation was achieved without impairing performance or availability. Closely integrated Power server system and workload management facilities were exploited to enable organizations to realize high levels of capacity utilization, while minimizing risks that partition workloads on the same server would interfere with each other, or cause performance bottlenecks or outages.

Full time equivalent (FTE) staffing levels for system administration-related functions such as asset, capacity, change, configuration and performance management are significantly lower for Power virtualized scenarios. Personnel costs for these functions are correspondingly less.

This is due to reductions in numbers of physical servers; reduced diversity of hardware and software platforms; replacement of older-technology servers and systems software with latest-generation Power hardware and software; improved management tools and practices, including increased automation; and use of PowerVM virtualization capabilities.

If costs of acquiring new Power server hardware and systems software to realize these savings were allowed for, overall costs were still significantly lower. Five-year costs ranged from 46 to 59 percent less, and averaged 54 percent less than those for conventional scenarios.

Figure 2 summarizes these results.

Figure 2 Conventional and Power Virtualized UNIX Server Scenarios: Five-year Overall Costs

(Virtualized Scenarios include Operating and Acquisition Costs)

In Power virtualized scenarios, use of Virtual I/O Servers resulted in I/O-related costs for servers employing LPARs that averaged 73 percent less than would have been the case if dedicated adapters were employed. Figure 3 summarizes these results.


Figure 3 Five-Year I/O-related Costs for Partitioned IBM Power Servers:

Averages for All Power Virtualized Scenarios

Savings were realized in acquisition, maintenance and support costs through use of fewer physical local area network (LAN) and storage area network (SAN) adapters, as well as in costs for related infrastructure components such as cabling, hubs and switches.

Costs of Downtime

Five-year costs of downtime for Power virtualized scenarios ranged from 43 to 54 percent less, and averaged 47 percent less than those for conventional equivalents. Figure 4 summarizes these results.

Figure 4 Conventional and Power Virtualized UNIX Server Scenarios: Five-year Costs of Downtime

COMPANY FINANCIAL SERVICES MANUFACTURING RETAIL CONVENTIONAL SCENARIOS

Availability levels Five-year costs ($000)

99.5% – 99.94% 26,266.8

99.83% – 99.9% 46,165.3

98.7% – 99.9% 14,552.5

POWER VIRTUALIZED SCENARIOS


99.8% – 99.97% 12,148.3

99.9% – 99.96% 26,285.4

99.3% – 99.95% 7,453.0

In this presentation, availability percentages reflect hours of system-level outages relative to annual hours of operation of companies, or business areas within companies that are supported by specific applications.

Costs of downtime for the financial services company represent lost operating profit, meaning net revenue after deduction of personnel, occupancy, equipment and other overhead. For the manufacturing and retail companies, the comparable metric is lost gross profit, meaning profit net of cost of goods sold, but before deduction of selling, general and administrative (SG&A) and other expenses.

Key Power availability strengths that contributed to lower costs of downtime included reliability, availability and serviceability (RAS) features of Power hardware and the AIX operating system, and the IBM PowerHA for AIX clustered failover solution.

LPARs also assisted in avoiding planned outages by allowing users to upgrade or modify software without taking systems offline. Two newer capabilities made it possible to transfer partitions between servers if these needed to be shut down: (1) Live Partition Mobility involved no interruption of service, while (2) Live Application Mobility involved interruptions of typically no more than 20 seconds.

Power server system and workload management capabilities further reduced risks of bottlenecks and outages caused by workload spikes and operational errors.


Linux Server Infrastructures

IT Costs

In addition to the AIX operating system, Power servers may also run the major Linux distributions in native mode. To address this potential, comparisons were made for different sets of Linux-based applications for the three profile companies. Comparisons were based on mixes of business as well as Web and intranet serving, file serving, software development and other applications.

Two sets of scenarios were then compared. First, in x86 server scenarios, applications were deployed on one- to four-way Intel-based servers from Dell, HP, IBM and other vendors, with limited use of VMware and Xen virtualization tools. Second, in Power server scenarios, multiple Linux operating system instances were hosted on Power servers using PowerVM.

For x86 server scenarios, operating costs – primarily hardware maintenance, personnel and facilities –were calculated. For Power server scenarios, operating costs as well as acquisition costs for server hardware and PowerVM licenses were calculated. Figure 5 summarizes results.

Figure 5 x86 and Power Linux Server Scenarios: Average Five-year Costs for All Installations

Five-year operating costs for Power server scenarios ranged from 55 to 61 percent less, and averaged 58 percent less than those for x86 server scenarios. If acquisition costs are included in Power virtualized scenarios, five-year costs for these ranged from 50 to 57 percent less, and averaged 52 percent less.

Costs of Linux operating systems and databases were not included. Operating system costs would depend on the number of instances employed, which was the same for x86 as for Power server scenarios. Database costs were not included because Linux servers were not widely employed as database servers in the organizations upon which profile installations were based.

Factors causing lower costs for Power scenarios included high levels of consolidation and capacity utilization enabled by LPARs; lower staffing levels and personnel costs for system administration-related functions; and lower facilities costs due to use of significantly fewer physical servers.

Costs of Downtime

Costs of downtime were also compared for x86 and Power virtualized scenarios for Linux serving, although a different quantification method was employed.


Linux servers in the organizations upon which profile installations were based did not typically support business-critical systems for which downtime could be measured in the same manner as for UNIX server comparisons. It was thus decided to employ lost productivity values based on interruptions caused to application users.

Using this approach, results for costs of downtime were calculated as shown in figure 6.

Figure 6 x86 and Power Linux Server Scenarios: Five-Year Costs of Downtime

COMPANY FINANCIAL SERVICES MANUFACTURING RETAIL x86 SERVER SCENARIOS


99.45% – 99.93% 1,943.7

98.5% – 99.75 2,686.8

99.25% – 99.85% 2,086.2

POWER SERVER SCENARIOS


99.93% – 99.98% 275.3

99.85% – 99.95% 689.1

99.88% – 99.98% 452.0

Costs of downtime for Power server scenarios ranged from 74 to 86 percent less, and averaged 79 percent less than those for x86 server equivalents.

New UNIX System Deployments

Cost Comparisons

Five-year IT costs as well as costs of downtime were also compared for new UNIX system deployments employing HP Integrity, IBM Power and equivalent Sun platforms.

Three different composite installation profiles were employed for these comparisons. These were for deployments of database-intensive systems with high levels of virtualization in telecommunications, distribution and manufacturing companies with revenues of between $500 million and $12 billion.

Installations included business-critical as well as secondary applications supported by multiple high-end, midrange and small server platforms.

Three sets of scenarios were developed:

1. HP VSE scenarios were built around HP Integrity servers employing the HP-UX 11v3 operating system and the three forms of partitioning – (1) hardware-based nPars and software-based (2) vPars and (3) Integrity Virtual Machines – supported by the company’s Virtual Server Environment (VSE).

2. IBM PowerVM scenarios were built around IBM Power servers employing the AIX 6.1 operating system, along with PowerVM including LPARs and AIX WPARs.

3. Sun scenarios were built around multiple Sun server platforms employing the Solaris 10 operating system.

These included SPARC Enterprise M series servers employing the company’s hardware-based partitioning technology, dynamic domains; T series (CoolThreads) servers employing software-based LDoms; and Sun x64 servers. The Sun application-based partitioning technology, Solaris Containers and Zones, was employed on all three platforms.


For clustered failover, HP Serviceguard and IBM Power HA solutions were employed in HP and IBM scenarios respectively, and an equivalent third-party solution was employed in Sun scenarios.

Five-year IT costs were then calculated for these scenarios for each profile installation. Costs included acquisition of hardware and licenses for operating systems; virtualization tools; cluster solutions and Oracle 11g databases and management tools; hardware maintenance and software update subscriptions and support; and facilities costs for data center occupancy, power and cooling.

Database costs again reflected the change in Oracle pricing for POWER6-based servers in March 2009. Personnel costs for system administration and related functions are not included, as these would have been largely similar, at least for HP VSE and IBM PowerVM scenarios.

Results are summarized in figure 7.

Figure 7 New UNIX System Deployments: Five-year IT Costs

Five-year IT costs for PowerVM scenarios range from 38 to 50 percent less, and average 43 percent less than those for HP VSE equivalents; and range from 48 to 67 percent less, and average 60 percent less than those for Sun equivalents.

Lower IT costs for PowerVM scenarios are due to multiple factors. One is the higher granularity of IBM LPARs compared to Hewlett-Packard’s nPars and Sun’s dynamic domains.

Among the organizations that provided input for this report, “hard” partitioning technologies were preferred for business-critical production environments whose performance and availability requirements were more demanding than the norm. This was consistent with general industry experience.


HP and Sun employ hard partitioning technologies based on cell boards in Integrity and M series servers respectively. In the case of HP, nPars can be configured in increments of four Itanium2 processors (eight cores), corresponding to the cell board structure of larger Integrity models. nPars are not supported on smaller Integrity models with up to four processors, which are not cell board-based.

Sun dynamic domains can be configured in increments of one system board (cell board) with four processors (16 cores using quad-core SPARC64 processors), or one-quarter of a system board with one processor (four cores), along with one quarter of the board’s memory and I/O resources.

There are, however, limitations in the overall number of dynamic domains supported, which are illustrated in figure 8. Dynamic domains are supported only on M series servers.

Figure 8 Numbers of Hard Partitions Supported: HP and Sun UNIX Server Platforms

HEWLETT-PACKARD

Model rx2660 rx3600 rx6600 rx7640 rx8640 Superdome

Processors 1 – 2 1 – 4 1 – 8 1 – 16 1 – 64

nPars n/a n/a 1 – 2 1 – 4 1 – 16

SUN MICROSYSTEMS Model M3000 M4000 M5000 M8000 M9000 Processors 1 1 – 4 1 – 8 1 – 16 1 – 64

Dynamic domains 1 1 – 2 1 – 4 1 – 16 1 – 24

Granularity limitations affected configuration sizes for Integrity as well as M series servers. In comparison, firmware-based LPARs allowed use of smaller partitions – granularity of up to 0.1 of a core is supported – and Power server mechanisms for allocating and reallocating resources between these were also more effective than HP and Sun equivalents.

As a result, system-level capacity utilization for hard-partitioned applications and databases was higher for Power servers than for HP and Sun M series equivalents. Moreover, it was possible to employ LPARs on comparatively small servers. LPAR capability is supported on all Power models.

(It is occasionally argued that nPars and dynamic domains allow for “electrical isolation,” which increases reliability as well as security. In practice, however, users have not found availability levels to be lower for Power servers, and LPARs provide comparable partition isolation for security purposes.)

Sun configurations were also affected by the fact that the company’s LDoms partitioning technology, which enables software-based hosting of operating system instances, is supported only for the company’s T series servers.

Superior price/performance levels for Power servers also contributed to lower costs for PowerVM scenarios The Itanium2 processor architecture is aging, and lags far behind POWER6 in all major industry performance ratings. Sun’s SPARC64 is generally recognized to be an even weaker design, delivering even lower performance relative to Power servers for comparable configurations.

Higher system performance, as well as higher capacity utilization enabled by PowerVM, offset what might otherwise have been a Power server disadvantage in database costs – Oracle license costs are calculated at 0.5 of a processor for Itanium2, 0.75 for SPARC64, and 1.0 for POWER6 processors.

For all profile installations, however, five-year Oracle database license and support costs were significantly lower for PowerVM than for HP VSE and Sun scenarios.


Finally, HP as well as IBM platforms benefited from integrated system and workload management capabilities that are, by wide margins, superior to those offered by Sun for Solaris servers. As a result, Sun platforms suffered a penalty relative to Integrity and Power servers in comparative capacity utilization levels.

Costs of Downtime

Costs of downtime for new UNIX system deployments were calculated in a manner similar to conventional and Power virtualized scenarios for UNIX server infrastructure comparisons.

Results are summarized in figure 9.

Figure 9 New UNIX System Deployments: Five-year Costs of Downtime

COMPANY TELECOMMUNICATIONS DISTRIBUTION MANUFACTURING HP VSE SCENARIOS


99.94% – 99.97% 5,497.0

99.97% – 99.98% 4,847.7

99.94% 1,338.1

IBM POWERVM SCENARIOS


99.96% – 99.98% 5,238.3

99.97% – 99.98% 4,568.9

99.95% 1,115.1

SUN SCENARIOS


99.93% – 99.97% 8,877.3

99.94% – 99.97% 6,377.1

99.92% 1,784.2

In this presentation, availability levels again reflect hours of system-level uptime relative to annual hours of operation of companies, or business areas within companies that are supported by specific applications. Costs of downtime again represent lost gross profit, meaning profit net of cost of goods and/or services, but before deduction of SG&A and other expenses.

Costs of downtime for the HP VSE and IBM PowerVM scenarios are generally similar. This reflects high levels of hardware RAS and strong clustered failover offerings, along with capabilities for partition-based software upgrades and maintenance, live partition movement between physical servers, and other availability optimization functions for both vendors’ platforms.

Marginally higher costs of downtime for HP VSE scenarios for reflect slightly higher levels of planned downtime for Superdome platform, as well as lower availability levels for secondary applications deployed on other Integrity servers.

Costs of downtime for Sun scenarios were significantly higher than for HP VSE and PowerVM scenarios. This reflects less effective hardware as well as software availability optimization across all Sun platforms. Live partition movement, for example, is supported only for LDoms on Sun T series servers, which were employed for secondary applications in the Sun telecommunications company scenario.


DETAILED DATA

UNIX Server Infrastructure Comparisons

Installations and Scenarios

Comparisons of five-year costs for conventional and Power virtualized scenarios for UNIX server infrastructures are based on the composite profile installations summarized in figure 10.

Figure 10 Profile Installations and Scenarios for UNIX Server Infrastructure Cost Comparisons

COMPANY FINANCIAL SERVICES MANUFACTURING RETAIL Business Profile $400 billion assets

$15 billion revenues 14 million customers 1,650 branches 55,000 employees

$33 billion revenues Consumer products 28 manufacturing plants 25 distribution centers 50,000 employees

$35 billion revenues Specialty chain 1,000+ outlets 10 distribution centers 150,000 employees

Focus of Comparisons

69 major applications System groups: • Corporate systems • Retail banking • Commercial banking • Financial services • Online systems • Intranet infrastructure

30 major applications System groups: • Core ERP systems • Supply chain management • Product management • Business intelligence • eProcurement • BPM, CRM, PLM

36 major applications System groups: • Corporate systems • Supply chain management • Logistics & transportation • Marketing, sales & service • Business intelligence • eCommerce

CONVENTIONAL SCENARIOS Servers Hewlett-Packard

Superdome, rx8640, rx8620, rx7620, rx6600, rx4640, rx3600, rx2660, rx2620, rx2600, rp8420, rp8400 IBM pSeries 690, 670, 650, 615 System p 595, 570, 550, 510, 505, Power 570, BladeCenter Sun Microsystems E25K, E6800, E4900, E4800, E2900, M4000, V890, V880, V490, V480, V40Z, X4600, X4500, X4200, X4100, X2000, Blade 6000, various Silicon Graphics Altix 450

Total: 242 servers

Hewlett-Packard Superdome, rx7640, rx7620, rx6600, rx4640, rx3600, rx2660, rx2620, rp8400, rp5470 IBM pSeries: 670, 650, 615 System p 570, 550, 520, 510 Power 570 Sun Microsystems E20K, E15K, V890, V880, V490, V40Z, X4600, X4200, X4100, X2000

Total: 78 servers

Hewlett-Packard rx8640, rx8620, rx7620, rx6600, rx4640, rx2660, rx2620, rx2600, rx1600 IBM pSeries 670, 650, 615 System p 570, 550, 520, 510, 505, 185 Sun Microsystems E15K, E4900, E4800, V890, V490, V480, V440, V250, V240, V40Z, V20Z, X4600, X4500, X4200, X2000, various Total: 98 servers

Personnel 28 FTEs 13 FTEs 14 FTEs

POWER VIRTUALIZED SCENARIOS Servers 9 x 570, 7 x 550, 3 x 520

3 x JS43, 16 x JS23 5 x JS12, 2 x BladeCenter H

Totals: 43 servers 125 LPARs 106 WPARs

5 x 570, 9 x 550 6 x JS23, 3 x JS12 BladeCenter H


5 x 570, 5 x 550 3 x 520, 1 x JS43, 14 x JS23 8 x JS12, 2 x BladeCenter H


Personnel 15 FTEs 8 FTEs 9 FTEs

Profile installations were constructed using data on applications, server bases, configurations, utilization and service levels, staffing and other variables supplied by 16 companies in the same industries and approximate size ranges, with generally similar business profiles.


Companies employed a variety of Hewlett-Packard, IBM, Sun and other servers employing HP-UX, AIX, Solaris and other UNIX variants.

Using this data, two sets of scenarios were developed:

1. Conventional scenarios are based on data reported by the 16 companies, and are built around diverse multivendor server bases that include different technology generations and systems software versions. Conventional management and operating practices are employed.

2. Virtualized scenarios are for the same applications and workloads deployed on latest-generation Power servers. Configurations for these scenarios were developed on a case-by-case basis within major system groups. Where appropriate, multiple applications were configured on the same physical servers using LPARs and/or WPARs.

In translating configurations employed in conventional scenarios into Power server configurations employed in virtualized scenarios, the following approach was employed.

Power server configurations delivering performance equivalent to conventional scenario servers were developed for each of the main applications in profile installations. Configurations were developed for production and non-production instances.

Further calculations were undertaken to allow for the effects of virtualization. Nominal configuration requirements were first determined for groups of instances were deployed on separate servers, but represented realistic candidates for consolidation onto a single Power physical server.

An overall utilization value reflecting realistic potential consolidation efficiencies was then assigned to each group of instances, and allowance was made for other factors affecting capacity requirements. The resulting configuration was then rounded to next largest capacity increment offered by IBM; e.g., a nominal configuration of 3.3 x 4.2 GHz cores with 12.6 GB of RAM was rounded to a 4 x 4.2 GHz 550 model with 16 GB of RAM.

Other hardware components were configured similarly. Allowance was made for use of other PowerVM capabilities, including Virtual I/O Servers and Integrated Virtual Ethernet.

All installations and scenarios include database, application and, where appropriate, Web and intranet servers; and production systems as well as non-production instances for functions such as development, test, quality assurance and training. Clustered failover configurations are employed in conventional and Power virtualized scenarios for systems requiring high levels of availability.

Financial services company scenarios include a variety of custom and packaged software solutions. Manufacturing company scenarios are built primarily around SAP and complementary third-party applications, including business process management (BPM), customer relationship management (CRM) and product lifecycle management (PLM) solutions.

Retail company scenarios include a mix of applications from EXE Technologies, i2 Technologies, JDA Software, Oracle (including PeopleSoft offerings) and other vendors. Costs of applications software were not included in calculations.

IT Costs

For conventional scenarios, calculations are for operating costs only. These include hardware maintenance, update and support subscriptions for systems and database software, along with personnel and facilities costs.


For Power virtualized scenarios, calculations are for operating costs, which include the same components, along with costs of hardware and systems software acquisition for new servers. Hardware maintenance costs are for 24x7 coverage.

For both sets of scenarios, systems software includes operating systems, virtualization software, clustered failover solutions and related tools. For Power virtualized scenarios, systems software includes IBM AIX 6.1, PowerVM, PowerHA and – for BladeCenter configurations – BladeCenter Open Fabric Manager.

Costs for database update and support subscriptions are for Oracle databases and management tools. Calculations do not include initial license costs. For Power virtualized scenarios, database costs are for Oracle 11g equivalents. Calculations reflect the change in Oracle pricing for POWER6-based servers announced in March 2009; i.e., calculations are based on a value of 1.0 processor.

Detailed breakdowns of IT costs are presented in figure 11.

Figure 11 Cost Breakdowns for UNIX Server Infrastructure Comparisons

COMPANY FINANCIAL SERVICES MANUFACTURING RETAIL

CONVENTIONAL SCENARIOS Maintenance 4,819.7 2,775.8 1,612.5

Systems software support 1,990.7 598.8 537.1

Database software support 13,803.7 5,733.0 5,051.1

Personnel 16,623.0 7,717.8 8,311.5

Facilities 1,173.1 548.4 432.3

TOTAL ($000) 38,410.2 17,373.8 15,944.7

POWER VIRTUALIZED SCENARIOS Hardware 2,025.5 1,345.1 1,086.3

Systems software 444.2 215.9 185.8

Acquisition costs (Subtotal) 2,469.8 1,561.0 1,272.2

Maintenance 461.4 251.1 215.6

Systems software support 1,122.0 303.8 234.2

Database software support 2,546.4 1,655.3 1,443.5

Personnel 8,905.2 4,749.4 5,343.1

Facilities 275.7 163.4 121.3

Operating costs (Subtotal) 13,310.8 7,122.9 7,357.8

TOTAL ($000) 15,780.6 8,683.9 8,629.9

Additional detail on the basis of calculations for personnel and facilities costs for these and other comparisons may be found at the end of this section.


Costs of Downtime

Costs of downtime for UNIX server infrastructures were calculated using industry- and company-specific values assuming consistent levels of availability and business cost structures over a five-year period.

For the financial services company, costs of downtime consists primarily of lost transaction fees and customer lifetime value (CLV) for outages affecting online banking, brokerage and other customer self-service systems, along with CRM systems and Internet infrastructure servers supporting these.

Costs of downtime for the manufacturing company were calculated based primarily on outages affecting the company’s core SAP ERP, supply chain management and procurement systems. For the retail company, costs were calculated based on outages affecting the company’s supply chain systems as well as Internet marketing, sales and customer service systems.

The impact of outages on the retail company’s Internet systems includes the effects not only of lost online sales, but also of storefront sales lost because customers are unable to obtain information on products, promotions, pricing, availability, store locations and other subjects.

Linux Server Infrastructures


Comparisons of five-year costs for x86 and Power Linux server infrastructures are based on the composite profile installations and scenarios summarized in figure 12.

Figure 12 Profile Installations and Scenarios for Linux Server Infrastructure Cost Comparisons

COMPANY FINANCIAL SERVICES MANUFACTURING RETAIL Applications Market value & risk analytics

Equities trading, market data Mortgage loans, antifraud Departmental applications Content management Software development & test Intranet applications Intranet infrastructure, various

SAP xApps Computer aided design Departmental applications Software development & test File serving Web infrastructure, various

Internal portal, e-procurement Promotional e-mail Departmental applications Office applications POS software development File serving, fax serving Network management Web infrastructure, various

x86 SCENARIOS Servers Dell, HP, IBM

128 physical servers 56 VMs

HP, IBM


Dell, HP, IBM, various


Personnel 5.1 FTEs 1.05 FTEs 2.55 FTEs

POWER VIRTUALIZED SCENARIOS Servers 3 x 550 4/8 x 4.2 GHz

4 x 520 2/4 x 4.7 GHz 185 LPARs

2 x 520 2/4 x 4.7 GHz 44 LPARs

2 x 550 4/8 x 4.2 GHz 120 LPARs

Personnel 2.15 FTEs 0.45 FTE 1.25 FTEs

The same companies are employed as for UNIX server infrastructure comparisons. Installation profiles, however, are different. Application portfolios, server bases and staffing levels were developed using data from 23 companies, including companies that, in some cases, did not supply UNIX server data.


These companies employed x86 servers from Dell, HP (including Compaq), IBM and other vendors. Bases included different hardware generations – ranging from Pentium 4-based servers installed in 2002 to recent Intel Xeon- and Advanced Micro Devices (AMD) Opteron-based servers – running Red Hat, SUSE and other Linux distributions.

x86 server scenarios are based on user-reported data. Power server scenarios were developed, and configurations sized in the same manner as for Power virtualized scenarios for UNIX server infrastructure comparisons. PowerVM Standard Edition is employed on all Power servers.

It is assumed that Linux applications are deployed on Power servers in native mode in a manner that effectively exploits the potential of PowerVM and that management and operating practices are supportive of this potential.

IT Costs

For x86 server scenarios, calculations are for operating costs only. These include hardware maintenance, and update and support subscriptions for VMware and Xen software, personnel and facilities costs.

For Power server scenarios, calculations are for operating costs, which include the same components, along with costs of hardware acquisition and PowerVM licenses for new servers. Operating costs include 24x7 hardware and software maintenance.


Figure 13 Cost Breakdowns for Linux Server Infrastructure Comparisons

COMPANY FINANCIAL SERVICES MANUFACTURING RETAIL x86 SERVER SCENARIOS Maintenance 190.1 42.6 85.1

Virtualization software 20.6 4.1 6.2

Personnel 2,721.5 560.3 1,360.7

Facilities 225.2 82.4 154.5

TOTAL ($000) 3,157.4 689.4 1,606.5

POWER SCENARIOS Hardware 232.0 26.0 92.8

Maintenance 50.8 8.2 20.3

Virtualization software 46.0 9.2 18.4

Personnel 1,147.3 240.1 667.0

Facilities 49.5 10.5 11.5

TOTAL($000) 1,525.6 294.0 810.0

For reasons discussed earlier, calculations do not include acquisition or ongoing costs for Linux operating systems and databases.


Costs of Downtime

Costs of downtime for Linux server infrastructure comparisons are for user productivity loss.

This was measured in terms of (1) idle time, reduced productivity or both for periods when applications were not available to users during working hours and (2) reduced productivity following an outage; e.g., a one-hour outage causing a 60 percent productivity reduction might be followed by a two-hour period in which a 20 percent productivity loss occurs.

Productivity loss calculations were developed for each user community supported by Linux applications. Costs are based on industry median salaries, benefits and other compensation for large U.S. financial services, manufacturing and retail companies for the occupational groups using applications.

Productivity losses were quantified based on the number of individuals using each application who would be affected by outages, their average remuneration per hour, and percentage values for reductions in their productivity during and subsequent to outages.

New UNIX System Deployments


Comparisons of five-year costs for new UNIX system deployments are based on the composite profile installations and scenarios summarized in figure 14.

Figure 14 Profile Installations and Scenarios for New UNIX System Deployments Comparisons

TELECOMMUNICATIONS COMPANY BUSINESS PROFILE APPLICATIONS

$12 billion revenues 20+ million customers Fixed line, mobile, broadband & data network services 20,000+ employees

CRM, billing, financial & operational systems, data warehouse, application servers, identity management, intranet, customer portal, development & test

CONFIGURATIONS HP VSE SCENARIO IBM POWERVM SCENARIOS SUN SCENARIOS

2 x Integrity Superdome 64/128 x 1.6 GHz each 28 nPars + 80 vPars rx8640 12/24 x 1.6 GHz 2 nPars + 15 Integrity VMs rx7640 8/16 x 1.6 GHz 10 Integrity VMs 2 x rx6600 4/8 x 1.6 GHz each 20 Integrity VMs HP-UX 11v3 VSE Serviceguard (Superdomes) Oracle 11g, Enterprise Manager

2 x Power 595 24/48 x 5.0 GHz each 110 LPARs 4 Virtual I/O Servers Power 550 4/8 x 5.0 GHz 2 LPARs + 15 WPARs Virtual I/O Server Power 550 4/8 x 4.2 GHz – 10 WPARs 2 x Power 520 2/4 x 4.2 GHz – 20 WPARs AIX 6.1, PowerVM PowerHA (Power 595s) Oracle 11g, Enterprise Manager

2 x Enterprise M9000 56/224 x 2.52 GHz each 28 dynamic domains 52 Zones 2 x T5440 4/32 x 1.4 GHz each 25 LDOMs 3 x T5140 2/16 x 1.2 GHz each 20 LDOMs Solaris 10 Third-party clustered failover (M1000e) & system management tools Oracle 11g, Enterprise Manager

Profile installations were constructed using data on applications, server and partition configurations, utilization and service levels, and other variables supplied by 15 companies in the same industries and approximate size ranges, with generally similar business profiles.


Figure 14 (Continued)

DISTRIBUTION COMPANY BUSINESS PROFILE APPLICATIONS

$7 billion revenues Chemical products distributor 40 distribution centers 10,000 employees

ERP, CRM, order management, warehouse management, transportation management, business intelligence, procurement

CONFIGURATIONS HP VSE SCENARIO IBM POWERVM SCENARIO SUN SCENARIO

2 x Integrity Superdome 36/72 x 1.6 GHz – 4 nPars 30 vPars & Integrity VMs 32/64 x 1.6 GHz 6 nPars + 20 vPars 2 x rx7640 8/16 x 1.6 GHz – 2 nPars 8/16 x 1.6 GHz – 12 Integrity VMs HP-UX 11v3 VSE Serviceguard (Superdomes) Oracle 11g, Enterprise Manager

2 x Power 570 12/24 x 4.2 GHz – 36 LPARs 2 Virtual I/O Servers 8/16 x 4.7 GHz 28 LPARs 2 Virtual I/O Servers 2 x Power 550 4/8 x 4.2 GHz – 2 LPARs 4/8 x 3.5 GHz – 12 WPARs AIX 6.1, PowerVM PowerHA (Power 570s) Oracle 11g, Enterprise Manager

2 x Enterprise M8000 16/64 x 2.52 GHz 8 dynamic domains 16/64 x 2.4 GHz 4 dynamic domains + 31 Zones Enterprise M5000 8/32 x 2.4 GHz 2 dynamic domains Sun Fire x4440 4/16 x 2.5 GHz – 13 Zones Solaris 10 Third-party clustered failover (M8000s) & system management tools Oracle 11g, Enterprise Manager

MANUFACTURING COMPANY BUSINESS PROFILE APPLICATIONS

$500 million revenues Food & beverage products 6 manufacturing plants 35 distribution facilities 1,800 employees

ERP, supply chain management, HRMS, business intelligence, development & test

CONFIGURATIONS HP VSE SCENARIOS IBM POWERVM SCENARIO SUN SCENARIO

rx8640 12/24 x 1.6 GHz 2 nPars + 6 vPars rx7640 8/16 x 1.6 GHz 2 nPars + 8 vPars rx6600 4/8 x 1.6 GHz – 9 Integrity VMs rx2660 2/4 x 1.66 GHz – 3 Integrity VMs HP-UX 11v3 VSE Serviceguard (rx8640 & rx7640) Oracle 11g, Management tools

2 x Power 550 4/8 x 5.0 GHz 10 LPARs 2 Virtual I/O servers 4/8 x 4.2 GHz 10 LPARs 2 x Power 520 2/4 x 4.7 GHz – 9 WPARs 1/2 x 4.2 GHz – 3 WPARs AIX 6.1, PowerVM PowerHA (Power 570 & 550) Oracle 11g, Management tools

Enterprise M8000 8/32 x 2.52 GHz 2 dynamic domains + 7 Zones Enterprise M5000 8/32 x 2.4 GHz 2 dynamic domains + 9 Zones Sun Fire x4240 2/8 x 2.7 GHz – 10 Zones Sun Fire x4140 1/4 x 2.7 GHz – 4 Zones Solaris 10 Third-party clustered failover (M8000 & M5000) & system management tools Oracle 11g, Management tools

Companies employed HP Integrity, IBM Power and System p, or Sun Microsystems SPARC-based, T series and x86 servers running HP-UX, AIX and Solaris operating systems respectively. All companies employed Oracle databases.


Where companies employed older versions of platforms, configurations were updated to current-generation models using vendor or industry comparative performance data. Comparable configurations were then developed of the other two vendors’ platforms based on industry comparative performance data and partition granularity; e.g., if an IBM LPAR was sized to utilize 1.2 POWER6 cores, this would be translated into the next-largest HP nPar or Sun dynamic domain increment.

A similar approach was employed for partitions utilizing shared resources. Comparative sizing was based on the overall number of processors and amount of memory in the resource pool.

Where this approach was more cost-effective than use of dedicated adapters, Power servers are equipped with one or (for redundancy purposes) two Virtual I/O Servers running in dedicated LPARs.

Configurations were equipped with operating systems, virtualization tools and, where appropriate, clustered failover solutions. Software for Integrity servers included HP Virtual Server Operating Environment (VSE-OE) Version 4.1, which includes HP-UX 11v3, and Serviceguard. Power servers were equipped with AIX 6.1, PowerVM Standard or Enterprise Edition as appropriate, and PowerHA.

Sun servers were equipped with Solaris 10, which includes support for dynamic domains on M series servers, LDoms on T series servers, and Solaris Containers and Zones on these and x86 servers; and a widely used third-party clustered failover solution. Logical Domains 1.1 was employed on T series servers.

Database servers were equipped with Oracle 11g Enterprise or Standard Edition, and Oracle Enterprise Manager (for the telecommunications and distribution companies), or Diagnostics and Tuning Packs (for the manufacturing company).

Telecommunications company scenarios include use of Amdocs, SAP, SAS, Siebel and proprietary systems; distribution company scenarios include ERP, CRM and operational systems supporting multiple business units; and manufacturing company scenarios include JD Edwards ERP and CRM systems. Costs of applications software are not included in calculations.

Scenarios include development, test, and other non-production and production instances. Non-production instances are in most cases deployed on failover servers.

IT Costs

For all installations and scenarios, calculations include acquisition, as well as hardware maintenance, update and support subscription, and facilities costs. All hardware maintenance and software support costs are for 24x7 coverage. For reasons discussed earlier, personnel costs are not included.

Calculations for Oracle database and management tool costs again reflect the company’s change in pricing for POWER6-based servers announced in March 2009; i.e., calculations are based on a value of 1.0 processor.



Figure 15 Cost Breakdowns for New UNIX System Deployment Comparisons

COMPANY TELECOMMUNICATIONS DISTRIBUTION MANUFACTURING HEWLETT-PACKARD VSE SCENARIOS Hardware 5,234.4 2,971.5 592.4

Maintenance 2,709.7 1,444.5 111.1

Systems software 1,164.8 753.5 197.2

Database software 6,552.0 3,822.0 1,186.5

Facilities 304.3 224.0 55.6

TOTAL ($000) 15,965.2 9,215.5 2,142.8

IBM POWERVM SCENARIOS Hardware 3,159.4 1,195.3 174.9

Maintenance 891.7 141.7 20.8 Systems software 682.9 995.9 89.8

Database software 4,914.0 2,184.0 830.6

Facilities 174.9 120.4 33.8

TOTAL ($000) 9,822.9 4,637.3 1,149.9

SUN SCENARIOS Hardware 7,588.1 2,726.4 627.5

Maintenance 591.1 276.5 110.1

Systems software 435.4 238.8 70.5

Database software 17,199.0 5,460.0 2,610.3

Facilities 501.6 281.3 102.1

TOTAL ($000) 26,315.2 8,983.0 3,520.5

Costs of Downtime

Costs of downtime for new UNIX system deployment comparisons were determined in the same manner as for UNIX server infrastructures, using industry- and company-specific values assuming consistent levels of availability and business cost structures over a five-year period.

For the telecommunications company, costs of downtime were calculated for outages affecting billing, CRM, financial and operational systems, as well as customer-facing online applications. Calculations include allowance for lost sales and CLV due to interruptions in call center as well as online customer service.

For the distribution company, costs of downtime were calculated for outages affecting ERP, CRM, order management, warehouse management, transportation management and procurement systems. For the manufacturing company, the focus was on outages affecting the company’s ERP and CRM systems. Allowance was also made in both cases for lost sales due to customer service interruptions.


Cost Assumptions

Server Costs

For all comparisons, costs for server hardware acquisition and maintenance, and for software licenses, and update and support subscriptions were calculated based on vendor list prices. These were discounted to reflect prevailing “street” prices for installed bases and new deployments comparable to those in profile installations.

Facilities costs include data center occupancy, power and cooling equipment and energy costs over a five-year period. Occupancy cost calculations were based on EIA 42U rack mount units and service clearances for these, plus allowance for inactive areas. A conservative assumption for annual cost per square foot for existing facilities was employed (i.e., costs do not include new facilities construction).

Costs for power and cooling equipment were based on configurations of such equipment appropriate for the servers employed in each installation and scenario. Costs were calculated for acquisition and maintenance over a five-year period, using discounted list prices for equipment from leading vendors.

Energy costs were calculated using vendor electricity consumption values for servers, as well as power and cooling equipment. Specific utilization levels and hours of operation for each profile installation were applied, and a conservative assumption for average price per kilowatt/hour was employed to determine five-year costs.

Personnel Costs

Personnel costs are for the numbers of full time equivalent (FTE) staff for system administration-related functions shown in figures 10 and 12.

Calculations were based on annual salaries of $79,316 for AIX system administrators, as well as for UNIX system administrators for other platforms for conventional scenarios for UNIX server infrastructure cost comparisons; and $71,292 for Linux system administrators for x86 and Power servers for Linux server infrastructure comparisons.

Salaries were increased by 49.7 percent to allow for benefits, bonuses, training and other personnel-related overhead, and overall costs were calculated for a five-year period.

All values for server as well as personnel costs are for the United States.

ABOUT THE INTERNATIONAL TECHNOLOGY GROUP ITG sharpens your awareness of what’s happening and your competitive edge

. . . this could affect your future growth and profit prospects The International Technology Group (ITG), established in 1983, is an independent research and management consulting firm specializing in information technology (IT) investment strategy, cost/ benefit metrics, infrastructure studies, deployment tactics, business alignment and financial analysis.

ITG was an early innovator and pioneer in developing total cost of ownership (TCO) and return on investment (ROI) processes and methodologies. In 2004, the firm received a Decade of Education Award from the Information Technology Financial Management Association (ITFMA), the leading professional association dedicated to education and advancement of financial management practices in end-user IT organizations.

The firm has undertaken more than 100 major consulting projects, released approximately 160 management reports and white papers, and delivered nearly 1,800 briefs and presentations to individual clients, user groups, industry conferences and seminars throughout the world.

Client services are designed to provide factual data and reliable documentation to assist in the decision-making process. Information provided establishes the basis for developing tactical and strategic plans. Important developments are analyzed and practical guidance is offered on the most effective ways to respond to changes that may impact or shape complex IT deployment agendas.

A broad range of services is offered, furnishing clients with the information necessary to complement their internal capabilities and resources. Customized client programs involve various combinations of the following deliverables:

Status Reports In-depth studies of important issues

Management Briefs Detailed analysis of significant developments

Management Briefings Periodic interactive meetings with management

Executive Presentations Scheduled strategic presentations for decision-makers

Email Communications Timely replies to informational requests

Telephone Consultation Immediate response to informational needs

Clients include a cross section of IT end users in the private and public sectors representing multinational corporations, industrial companies, financial institutions, service organizations, educational institutions, federal and state government agencies as well as IT system suppliers, software vendors and service firms. Federal government clients have included agencies within the Department of Defense (e.g. DISA), Department of Transportation (e.g. FAA) and Department of Treasury (e.g. U.S. Mint).

International Technology Group 4546 El Camino Real, Suite 230

Los Altos, California 94022-1069 Telephone: (650) 949-8410 Facsimile: (650) 949-8415 Email: [email protected]

ITG