Consolidating sap applications for improved operational costs looking at sap and linux running on ibm power systems

WHITE P APER
Consolidating SAP Applications for Improved Operational
Costs: Looking at S AP and Linux Running on IBM Pow er
Systems
Sponsored by: IBM

Randy Perry Jean S. Bozman
Aimee Bartlett Scott Guinn
December 2010
www.idc.com

EXECUTIVE SUMMARY
Enterprise applications support the daily operations of a business, which means that
they are vital to its ability to generate revenue and profit growth — and to account for
F.508.935.4015

it. Whether those applications support a supply chain, general accounting, personnel
management, or customer relationship management, they must be available to end
users and must be accessible enough to create "snapshots" of the business on a
daily, hourly, or, in some cases, near-real-time basis.
P.508.872.8200

IT customers have a choice of platforms on which to deploy these applications;
indeed, the applications may run on a mix of platforms throughout a datacenter or
across an organization. For organizations that rely on data to make business
decisions, the platform choices must include consideration of the operational
requirements and what it will take to deliver on the expected, and agreed-upon, levels
Global Headquarters: 5 Speen Street Framingham, MA 01701 USA

of service for those applications.

IDC studied a group of enterprises that had done just that: They had deployed SAP
applications on IBM Power Systems servers running Linux server operating
environments and had been working with those systems for several years. The study
was conducted with the aim of finding out the results of initial investments in those
platforms and how those investments had impacted operational costs over time. Key
findings from the study include:

Annual benefits totaled $25,000 per 100 users. For a large organization with
10,000 users, this would amount to more than $2 million in benefits.

Hardware savings amounted to $11,500 per 100 users or more than $1.1 million
per 10,000 users.

The utilization rate for servers more than doubled, from 34% to 70%.

Over a three-year period, companies saw discounted benefits of $59,967 per 100
users and an average investment of $13,625 per 100 users, resulting in a return
on investment (ROI) of 340% and a payback period of 6.7 months.

The results from the IDC study would be expected to be amplified if the study
were repeated with POWER7 based systems because of the enhancements in
POWER design that improve efficiency compared with POWER6 based systems.

SITUATION OVERVIEW
The economic downturn has greatly influenced how IT organizations and senior
business managers view IT spend. The business value of that IT spend must be
demonstrated in order to be justified and paid for. And yet, if current infrastructure is
not evaluated and vetted for operational costs, maintenance of the status quo in the
datacenter can also prove to be costly — even if no new systems are added.

That is why 2009 began an intense period of study and evaluation in which IT
managers looked at servers, storage, software, and services with an eye to keeping
overall capex costs in check while driving operational costs down. In many cases,
new IT purchases were kept to a minimum, while the study of what was already
in-house continued. Some of the results of this "inventory" process were surprising
with regard to the rapid growth of small servers, which individually kept server
purchases under the capex limits set by policy but collectively were causing
operational costs to rise, especially with respect to maintenance, IT staff costs, and
energy costs related to power and cooling.

This year, many organizations cautiously reentered the IT purchasing phase but did
so only if the avoidance of future hardware costs could be forecast and if existing
workloads could be consolidated into fewer server "footprints" in the datacenter. This
applied across the board — affecting all types of servers running all types of
operating systems.

Deploying Enterprise Applications in the
Datacenter

Many IT organizations have deployed enterprise applications across multiple "tiers" of
computing over time. This means that some are end-to-end applications — tapping
the Web-serving, application-serving, and database-serving "tiers" of the datacenter.
In other cases, specific enterprise application modules are deployed within a business
unit, such as those that support business unit analysis or accounts receivable.

The major enterprise applications, including those of SAP, are able to run on many
types of servers, hosted by many types of operating systems and processor
architectures. However, earlier trends toward highly distributed computing led to the
use of many dedicated servers, which ran just one application module, or the use of
multiple servers running the same application, for purposes of redundancy and
availability.

In many cases, this approach kept applications close to the end user, but it did not
take advantage of improvements in high-speed networking and the ability to
consolidate workloads onto more scalable servers for more efficient operation. Today,
customers have a wide range of deployment choices, and some have decided to
consolidate some — but not all — of these enterprise applications onto scalable
servers for central site management and to reduce ongoing operational costs
associated with maintenance, repeated software upgrades and security patches, and
the need to maintain large numbers of small servers, many of them distributed over
multiple sites.

2 #226173 ©2010 IDC

Reducing Operational Costs of Enterprise
Applications in the Datacenter

In recent years, IT organizations have developed a sharp focus on reducing
operational costs — including IT staff costs, management and maintenance costs,
and the costs of unplanned downtime, which can take important systems offline for
minutes, or hours, rendering those applications inaccessible for end users and end
customers within the supply chain. Energy costs associated with power and cooling,
which coincidentally increased during the recent economic downturn, are another
category of operational cost that has come under increasing scrutiny in recent years
given cost constraints across the business and limits on IT budgets.

IDC research has shown that rapidly increasing power and cooling and maintenance
costs have been top of mind for IT managers and datacenter managers — prompting
a cycle of evaluating current IT infrastructure to improve overall efficiencies, including
the use of datacenter floor space and energy efficiency. The adoption of virtualization
has helped in this regard, with many IT organizations leveraging virtualization to be
able to consolidate workloads onto fewer server footprints. But server design has also
changed, with multicore, multiprocessor systems playing an important role in
supporting workload consolidation and improved operational efficiency. Importantly,
the ability to manage the physical servers — and the virtual servers that are hosted
by them — has proven to be a key component of improving operational efficiency.

Consolidating Workloads on Scalable Servers
in the Next-Generation Datacenter

IDC's most recent annual Server Workloads and Server Virtualization studies have
produced demand-side, customer-based data that clearly demonstrates the steady
march toward improved efficiency in the datacenter. In fact, IDC research shows that
2010 was the "crossover" year in which the number of virtual servers exceeded the
number of physical servers shipped (see Figure 1), demonstrating the business value
of virtualization and consolidation of workloads.

©2010 IDC #226173 3

FIGURE 1

Server Virtualization Crossover in 2010

18,000,000
16,000,000
14,000,000
12,000,000
10,000,000
8,000,000
6,000,000
4,000,000
2,000,000
0
2005 2006 2007 2008 2009 2010 2011 2012

Physical servers Virtual machines

Source: IDC, 2010

What's driving this consolidation is economics; hardware investments can be more
highly utilized than before, and the workloads can still be isolated from each other to
prevent unplanned downtime or contention for resources (processors, memory, I/O).
Consolidation and virtualization are increasingly combined, across the board, whether
on x86 servers, RISC servers, or mainframes. At the same time, virtualization
management, often involving automation, has become essential for keeping
operational costs low.

It may be surprising to many to discover that a large percentage of all business-
critical and mission-critical applications and databases are running on scalable
servers. These larger servers offer in-the-box capacity on demand, advanced
virtualization, and highly granular control of all resources under management.

As defined by IDC, business processing workloads running on servers today include
online transaction processing (OLTP), ERP, and CRM. These are mission-critical
workloads, which are highly demanding of computing resources. As seen in Figure 2,
a large portion of business processing workloads today (34%) run on RISC server
platforms — a type of server architecture on which the IBM Power Systems servers
are built. Figure 2 also shows the other major types of platform architectures: EPIC,
x86, and CISC (which is the basis for scalable host servers such as mainframes).
Business processing is one of seven major workloads types studied by IDC on an
annual basis, based on surveys of more than 1,000 customer sites. The other major
workload types are decision support (analysis of data), collaborative (email and
groupware), application development, IT infrastructure (e.g., support of networking
protocols and file/print), Web infrastructure (e.g., Web serving, proxy, and caching),
and industrial R&D (e.g., technical computing and high-performance computing).

4 #226173 ©2010 IDC

FIGURE 2

IDC's Server Workloads Revenues by CPU Type, 2009

Business processing

Decision support

Application development

Collaborative

IT infrastructure

Web infrastructure

Industrial R&D

Other

0 2,000 4,000 6,000 8,000 10,000 12,000 14,000
($M)

x86 EPIC RISC CISC

Source: IDC's Server Workloads Multiclient Study, 2010

By consolidating workloads onto scalable servers, customers gain the advantages of
more computing resources, more capacity, improved management of workloads, and
more efficient use of real estate space and power than would be the case with larger
numbers of smaller servers distributed across the datacenter or multiple sites.
Although it cannot be suggested that all workloads could, or would, be fit onto just a
few scalable servers, it is possible to identify mission-critical applications that would
benefit from running on a scalable, highly available server platform.

IBM'S POWER SYSTEMS SOLUTIONS
The next-generation datacenter is all about capability — capability to scale up,
capability to be energy-efficient, and capability to host important workloads with high
reliability and availability. Due to the operational characteristics of IBM Power
Systems, some customers have studied ways to consolidate workloads onto more
scalable servers — even if they are accessed from multiple remote sites.

The business benefits can be impressive, with some aspects of operational costs
bring dropped dramatically compared with earlier scale-out deployments.

©2010 IDC #226173 5

SAP on Linux Deployments: Contributing
Factors Cited by Respondents

For the SAP customers in this IDC study, leveraging Linux is an important way to
optimize their application deployment environment, making it easier to work with other
types of open source software for ease of integration. Many of them have had long
experience in working with Linux, whether because of longtime use or availability of
Linux expertise in their geographic market.

Importantly, access to a deep pool of Linux programmers and developers was often
cited by the study's respondents as a driver for choosing a Linux-based platform.
Finding and retaining programmers
with deep Unix skills is often Deployment Scenarios for SAP with Linux on IBM Power Systems
expensive and difficult, compared
with the broader group of
IBM is now shipping Power Systems based on POWER7 processors, refreshing
developers with Linux skills, many its entire line of Power Systems servers in calendar year 2010. These servers
of the respondents said. range from small entry servers in the Power Systems Express series to midrange
and high-end servers that offer scalability and advanced virtualization,
partitioning, and management capabilities.
In this study, all of the customers
were running SAP on their POWER7 processors bring faster performance, dramatically improved
performance/watt energy efficiency, and up to eight cores per POWER7 multicore
enterprise systems. In some processor. POWER7 supports more scalable, highly virtualized systems, with
cases, they ran it across multiple, granularity of as little as 1/10 of one processor devoted to a single workload.
smaller servers within the business Virtualization Enables Consolidation
unit, but in other cases, IBM Power Systems servers have built-in virtualization capabilities through IBM
consolidated SAP workloads found PowerVM hypervisor technology. They can be highly virtualized, with up to 10
their way to the IBM Power micropartitions per POWER processor. Within the partitions, customers have the
option to run one of three operating systems: IBM AIX Unix, Linux (Red Hat
Systems running IBM AIX Unix or,
Enterprise Linux or Novell SUSE Linux), and IBM "i" (the follow-on product to i/OS
in this case, Linux. or AS/400 for IBM System i servers). By isolating these workloads within partitions,
PowerVM supports workload consolidation on a Power Systems server.
A key driver for this deployment Consolidation of workloads leverages these virtualization capabilities, allowing
decision, customers explained, workloads to reside side by side within the same processor or partition — or
was economy of scale in deploying within different partitions within a Power Systems server. Finally, advanced on-
board system management software directs the workload to available hardware
SAP on a large machine, where
resources, dynamically reallocating workloads as processing requirements
multiple instances could be change and as business needs change over time.
managed more easily. Other
reasons included familiarity with the applications and operating environments and
achievement of high levels of availability by hosting the applications on a high-end
server with little to no unplanned downtime.

The centrality of SAP to many of the IT sites studied by IDC is made clear by this
comment from one customer: "Creation of revenue is not part of the SAP system [we
have implemented], but our entire billing, help desk management, CRM, inventory, all
those areas — the entire shop for us — is on SAP," he said. "It is extremely critical —
90% of our mission-critical operations are run either directly through SAP or association
with SAP."

Skill sets and familiarity with Linux were also factors in the decision to deploy SAP on
Linux for these Power Systems, although these systems support three types of
operating environments: IBM AIX Unix, Linux, and IBM "i," which supports
applications that once ran on IBM AS/400 and IBM System i. For some, Linux has

6 #226173 ©2010 IDC

already been selected as a cross-platform operating system that spans the
enterprise. "Linux is a strategic server platform [at our company]," said one IT
manager, "because it supports open and vendor-neutral architecture and deploying it
brings cost benefits."

In many cases, the familiarity came with working with Linux on x86 server systems first
— prior to selecting Linux on Power. "We have a technical background and our
colleagues have a
background in Linux running Implementation Trends of SAP
databases and other
applications," said one IT
IDC sees ERP customers, amid the backdrop of an economic environment that has
manager. For this customer, brought extraordinary levels of board-level oversight to IT budgets, continuing to look to
three Power Systems drive costs and risks out of their ERP upgrades. SAP and ERP vendors in kind are
servers are supporting the responding by going beyond application delivery to provide even greater value for their
clients in areas such as advisory and transformation services. For advisory services, this
Linux workloads that were means offering upgrade assessments, business process standardizations, and
running on more than 15 architecture assessments. Transformation services come by way of enterprise application
dedicated servers — with integration and process modeling, application upgrades, platform migration, or architecture
and infrastructure transformations.
each server devoted to one
SAP application — in the For customers, a key driver for consolidation of multiple ERP instances as part of the
upgrade or reimplementation process is the desire to transform to a shared services
datacenter. platform among business units for back-office functions such as HR, financials, and
procurement, gaining significant TCO reductions in the process. Further efficiencies are
Consolidation of workloads gained by reducing the number of ERP suppliers, which will help pare the company's IT
brought important savings skills requirements and ease integration challenges. Streamlining the ERP datacenter and
supporting infrastructure has similar benefits, not only reducing overhead and lowering
by reducing operational TCO but also providing greater flexibility for adjusting to spikes in workload demands.
costs, which the tables in
SAP is a suite of modular applications for the enterprise. Originally developed for use on
this paper show. These the IBM mainframe, SAP was ported to many other operating systems over the years.
operational cost reductions, Today, it runs on all major computer architectures and all major operating systems,
as well as cost avoidance, including Unix, Linux, and Microsoft Windows.
are critical to explaining how In the enterprise, scalable SAP systems tend to be sized to the number of end users
the initial investments were they support, but the actual deployment is still shaped by customer preferences. So the
choice of platform and operating system ultimately comes down to customer preference,
recouped in just over a half-
as well as performance.
year of deployment.
SAP has run on Linux since the late 1990s, making it a proven, well-described computing
environment. Multiple Linux distributions are supported, but the main ones for Power
The savings included Systems are Novell SUSE Linux Enterprise System (SLES) and Red Hat Enterprise Linux
notable reductions on (RHEL). Both of these Linux distributions have been designed for use with enterprise
software licensing costs due workloads and have built-in support for high availability, security, and demanding
workloads that are accessed by hundreds to thousands of end users.
to the lower number of
instances of IBM software SAP was originally developed for use on IBM mainframes, but that version (R2) was ported
to Unix systems in the 1990s and later was ported to Linux and Microsoft Windows. Today,
products (e.g., Tivoli
customers can run SAP modules across their enterprise computing environment, but the
management) or third-party characteristics of the platforms that run SAP applications may determine the deployment
(e.g., Oracle database) patterns for specific instances.
licenses needed to support In the case of Power Systems, overall system scalability; built-in, advanced virtualization;
the consolidated SAP and the ability to run both the SAP modules and the database that stores the application
environment on Power data on the same machine are factors that could lead to a decision to consolidate
workloads. Frequently, SAP modules are deployed within business units, often running on
Systems. Some customers Windows or Linux systems, but they may not be colocated at, or even nearby, the central
cited this as a major factor in datacenter site for the company. The IT managers in this study found it advantageous to
their decision to consolidate colocate multiple SAP applications on fewer "system" footprints, reducing management
costs, licensing costs, and IT staff costs in many cases.
SAP on Power Systems.

©2010 IDC #226173 7

KEY FINDINGS OF THE IDC ROI STUDY
IDC's business value methodology is based on quantifying the experiences of end-
user organizations and then translating those experiences into a financial analysis
using a standard approach and assumptions as well as commonly accepted financial
metrics and formulas. To make the results relevant to all types of organizations, we
report them in terms of a common denominator — in this case, per 100 users. The
details regarding the key findings can be found in the tables that follow.

The Customers

IDC studied a group of enterprises that have deployed SAP applications on IBM
Power Systems servers and have been working with those systems for several years.
The study was conducted with the aim of finding out the results of initial investments
in those platforms and how those investments impacted operational costs over time.

Specifically, IDC conducted in-depth interviews of 10 organizations, each of which
deployed SAP applications on IBM Power Systems (using the IBM POWER6
processors) running Linux. These companies ranged in size from 350 to 5,000
employees; they were based in a variety of geographies, including the United States,
Europe, and Asia/Pacific. They represented many vertical industries, including retail,
financial services, manufacturing, and energy. On average, these companies had
1,900 employees, with an IT staff that numbered in the hundreds, and they had 600+
physical servers under management. Most were running multiple SAP modules,
including those for human resources (HR), ERP, and CRM.

The interviews yielded information defining the organizations' up-front investment costs
in the technology, deployment, and ongoing maintenance of the systems. The
interviews elicited the companies' experience with tangible and measurable IT benefits
and end-user business benefits over a three-year period. IDC's Business Value team
combined all of these factors in the synthesis of an overall ROI calculation.

Each site managed a unique inventory and mix of server platforms. Most of these
organizations were large companies in which the server infrastructure supported tens of
thousands of intraenterprise end users — and an even larger number of extraenterprise
users, including end customers, accessing their enterprise systems. Each of these
organizations deployed IBM Power Systems for purposes of workload consolidation onto
fewer server "footprints," and each had used them over a period of three years or more.

THE BENEFITS OF CONSOLIDATION OF
S AP-LINUX ON POWER
IDC's analysis of the interviews revealed that the organizations in the study have
experienced the following benefits:

IT infrastructure cost reduction — reduced capex spending by $11,150 per 100
users annually or more than $1.1 million for a company of 10,000 users

8 #226173 ©2010 IDC

IT staff productivity increase — reduced the IT staff time spent on manual and
maintenance activities, freeing up resources for more business-related activities
and to expand SAP process automation applications without adding overhead by
$5,840 per 100 users annually or roughly $600,000 for a company of 10,000 users

End-user productivity increase — reduced SAP application downtime by 74%,
increasing overall user productivity by 1% and adding $7,355 per 100 users
annually or roughly $735,000 for a company of 10,000 users

Revenue increase (operational income) — reduced downtime related to
customer-facing SAP operations, increasing operational income by $676 per 100
users annually or roughly $67,000 for a company of 10,000 users

IT Staff and User Productivity

As Figure 3 shows, combined IT staff and user productivity increases accounted for
over half of total annual benefits.

FIGURE 3

Annual Benefits per 100 Users

Operational prof it
increase ($676)

User productivity
savings ($7,355)
IT inf rastructure
cost reduction
($11,150)

IT staf f
productivity
increase ($5,840)

Source: IDC, 2010

This increase in productivity affected IT staff as well as company employees (user
productivity). IDC calculates these forms of productivity separately, but both are the
result of building a more efficient platform for delivering the SAP applications. IT
productivity (see Table 1) comes from time savings as a result of consolidated
operations, higher quality of services, reduced maintenance, and the effects of reduced
downtime (which, in turn, reduces the need for additional IT staff time associated with
bringing the systems back online and repairing whatever damage has occurred).

©2010 IDC #226173 9

T ABLE 1

IT Staff Productivity

IT Staff Productivity % Savings Value per 100 Users ($)

Server installation 93 476

Data backup 57 655

Security (patching) 100 1,071

Help desk 62 1,797

Total IT staff savings 33 5,840

Source: IDC, 2010

User productivity (see Table 2) is associated with improvements in the availability of
the applications and in the ability to work. This effect of business continuity ripples
throughout an organization, allowing end users — and end customers — to access
applications and associated data, often on a 24 x 7 x 365 basis (around the clock).
Today's global business has drastically reduced the "window" for scheduled downtime
because end users and end customers around the world need to access the same
systems. This means that high levels of uptime for applications are essential to high-
performing organizations.

T ABLE 2

User Productivity/Service Quality

User Productivity/Service Quality Hours per User Value ($)

Downtime hours per year before 4.45 121

Downtime hours per year after 1.14 31

Downtime savings 12.77 hours (74%) 90

Help desk hours per year before 0.37 17

Help desk hours per year after 0.10 4

Help desk savings 1.06 (89%) 12

Overall user productivity increase from faster, more reliable SAP 1% 614
application operation

Total user productivity benefit per user 716

Source: IDC, 2010

10 #226173 ©2010 IDC

The amount of unplanned downtime experienced by the study sites was one-quarter
the amount they had reported before workload consolidation onto Power Systems.
This avoidance of downtime translates into cost avoidance, improved user
productivity, improved IT productivity, and more revenue to the business through
avoidance of system downtime. In general, IDC finds that the reduction of downtime
affects all aspects of a customer's business because it optimizes business operations
and business results and avoids costs associated with IT staff, such as help desk
activities, and improves user productivity (across the organization).

Creating a More Efficient Platform

As Table 3 shows, the average utilization rate per server rose from 34% to 70%
following consolidation of SAP workloads on Power Systems running Linux. Increased
resource utilization is a consequence of virtualization and consolidation, which result in
having the applications run more efficiently on the consolidated platform and, in the
process, avoiding the hardware costs by deferring the need to have additional capacity.

T ABLE 3

Key Operational Improvements

Key Operational Improvements Before After

Utilization rate (%) 34 70

Number of CPUs supporting SAP 24 9

Source: IDC, 2010

In the study group, the number of processors running the SAP workloads was
reduced from 24 to nine, improving the administration and management of the
hardware resources. Because the Power Systems are already virtualized through the
use of the PowerVM hypervisor on Power Systems, these customers have the option
to increase the number of processors, as needed, when new workloads are added to
the system. Further, new workloads that would eventually be added to the machine
could run on IBM AIX or Linux, depending on the nature of the workloads and such
factors as customer preference. This promoted customer choice for deployments.

The consolidated, more efficient platform enabled a reduction in hardware costs of 55%
(see Figure 4). This result reflects not only the immediate benefit of reducing server
hardware today but also the avoidance of future hardware purchases related to the
expansion of the SAP modules to support continued process automation. In most cases,
reductions in the current hardware environment also drove reductions in related IT costs
such as software (18% reduction), datacenter space (5% reduction), networking
infrastructure (4% reduction), and energy for power and cooling (18% reduction).

These cost savings, in combination with the increase in IT staff productivity, enabled
the companies to reduce their costs to deliver SAP business applications by 31%.

©2010 IDC #226173 11

FIGURE 4

IT Infrastructure Cost Reduction

Before

After

0 1,000 2,000 3,000 4,000 5,000 6,000
($)

IT staff Hardware Software Datacenter

Source: IDC, 2010

Increased Operational Income

Based on the respondent data, 40% of the organizations in the study recognized
revenue associated with SAP operations that were interfacing with their end
customers. By reducing application downtime by 52%, thereby improving time to
market and reliability, they were able to increase their top-line revenue by an average
of $1.7 million annually. For the purposes of combining revenue benefits with cost
savings as part of an ROI analysis, IDC converts revenue to operational income. IDC
does this by assuming that every $1.00 of revenue requires $0.90 in costs (10%
operating margin). In this analysis, $1.7 million in revenue translates to approximately
$170,000 in operational income benefit.

ROI Analysis

IDC's methodology for ROI analysis involves assessing the discounted cash flows
that were generated by the investment over a three-year period. Our standard
discount rate, which typically is the cost of capital invested, is 12%. This is sufficiently
high to cover most organizations' cost of capital, plus any additional risk associated
with higher than average investment or with delayed recognition of benefit. The
investment in consolidating SAP-Linux on Power platforms consists of any initial costs
for hardware and software, installation, consulting services and additional labor
associated with consolidation and migration, up-front training, annual support, and
maintenance. The consolidation process varied significantly because some
companies chose to stagger deployment over time; however, on average, companies
consolidating as one continuous process required 10 months to complete.

12 #226173 ©2010 IDC

Over a three-year period, companies saw discounted benefits of $59,967 and an
average investment of $13,625 per 100 users, resulting in an ROI of 340% and a
payback period of 6.7 months (see Table 4).

T ABLE 4

Three-Year ROI Analysis per 100 Users

Average

Average benefit (discounted) $59,967

Average investment (discounted) $13,625

Net present value $46,342

Discount rate 12%

ROI 340%

Payback period 6.7 months

Source: IDC, 2010

This short payback period may be surprising to some. Although Power Systems are
priced from less than $10,000 to more than $1 million per server, the midrange and
high-end models have generally gained the most attention. But due to highly efficient
virtualization and efficient operations, the Power Systems show attractive operational
costs when used over a three-year period. Given the overall costs examined in this
study and the detailed IDC analysis of actual costs seen by Power Systems sites, it
can be seen that operational costs were greatly reduced through the use of POWER6
based systems, which led to the conclusion that initial costs could be recouped within
a year. IDC notes that the use of POWER7 based systems, which began shipping in
February 2010, would be expected to further reduce operational costs and to improve
the average benefits and the ROI results shown in this study.

CHALLENGES/OPPORTUNITIES

Challenges

While this study has demonstrated that deploying SAP-Linux on Power Systems is
highly efficient, it is not the choice of all Power Systems customers. That is because
multiple deployment decisions can be made, depending on IT experience with the
specific computing environment under discussion (in this case, Linux on Power
Systems) and depending on the familiarity of programmers and system administrators
with that specific environment.

©2010 IDC #226173 13

This study found that SAP-Linux deployments on Power Systems brought notable
reductions in operational costs compared with earlier deployments at these customer
sites. However, other decisions could have been made that would also have brought
operational benefits, compared with older systems, such as running the same
applications under IBM AIX Unix, depending on customer preference and skill sets.

Customer preferences for operating environments and for hardware architecture often
are high of the list of selection criteria; therefore, some customers may lean toward
other choices, including deployment of SAP-Linux on x86 servers, especially if their
background has been deeply rooted in the x86 world. However, it is equally clear that
IT sites should fully explore the option of leveraging Power Systems given their
scalability, availability, support for security, and ability to consolidate multiple
workloads that may have been running on x86 servers in the datacenter.

Opportunities

The datacenter is being transformed, and many of the traditional deployment patterns
are undergoing change, prompted by economics and by an interest in reducing
operational costs. Because of this, workloads that once ran on many dedicated
servers — servers running only one application — are being consolidated to run on
fewer servers running consolidated workloads, often for the purpose of reducing
energy costs related to power and cooling and IT staff costs.

Taking an inventory of where applications are being run today and an IT "snapshot" of
current deployments could lead to changes in an IT site's future deployment patterns.
As this inventory is under way, factors to consider include the following: energy
efficiency, ease of use, IT staff time, and high availability for the enterprise applications
that run the business.

Customers have an opportunity to change their traditional patterns for deployment within
the datacenter or computer room. They can look at new deployment patterns that could
reduce the use of datacenter floor space as well as reduce energy costs and
maintenance costs. While there are many possibilities in this wave of datacenter
transformation, those themes will be a guide to next steps in building a more efficient
datacenter infrastructure.

CONCLUSION
IDC's study of organizations that had deployed SAP applications on IBM Power
Systems running Linux found that these sites experienced operational cost savings
over a period of three years following initial deployment. These companies, which
ranged in size from 350 to 5,000 employees, were based in a variety of geographies,
including the United States, Europe, and Asia/Pacific.

The study, which was based on in-depth interviews, determined the extent of customer
investments in existing Power Systems technology and the resulting IT and business
benefits associated with those systems. The IBM Power Systems servers studied were
considered to be "critical" to the businesses that deployed them. That means that
downtime would have affected both user productivity and company revenue.

14 #226173 ©2010 IDC

The systems, which were scalable, supported workload consolidation from larger
numbers of smaller servers.

By leveraging consolidation, and by combining it with virtualization, customers saw
efficiencies associated with hosting the SAP modules on IBM Power Systems.
Overall, the respondents reported that they saw measurable IT and business benefits
from these deployments, amounting to millions of dollars per year in savings for large
enterprises. In addition, they saw ROI payback periods of just over half a year
following initial investment in/acquisition of IBM Power Systems.

Copyright Notice

External Publication of IDC Information and Data — Any IDC information that is to be
used in advertising, press releases, or promotional materials requires prior written
approval from the appropriate IDC Vice President or Country Manager. A draft of the
proposed document should accompany any such request. IDC reserves the right to
deny approval of external usage for any reason.

Copyright 2010 IDC. Reproduction without written permission is completely forbidden.

©2010 IDC #226173 15

Consolidating sap applications for improved operational costs looking at sap and linux running on ibm power systems

More Related Content

What's hot (20)

Similar to Consolidating sap applications for improved operational costs looking at sap and linux running on ibm power systems (20)

More from IBM India Smarter Computing (20)

Recently uploaded (20)

Consolidating sap applications for improved operational costs looking at sap and linux running on ibm power systems