If your company is considering whether to consolidate its Intel servers, you should check out a new study from International Data Corporation (IDC) that IBM posted to its iSeries Web site late last month. The study, which IBM commissioned, claims that small and medium-sized businesses can achieve significant savings by consolidating Windows and Linux servers to the iSeries. While the study has its flaws, it does present useful information that iSeries professionals can incorporate into their server consolidation proposals.
In its study, IDC examined six companies that consolidated Intel servers to the iSeries' Integrated xSeries Server or to an iSeries logical partition running Linux. All six companies consolidated Windows on Intel servers to the iSeries, while two of the companies also consolidated Linux on Intel servers. On average, the six companies had 725 employees, annual revenues of $270 million, and an IT department with 11 staff members. The average company had 34 servers before its consolidation project and 20 servers afterward.
To determine whether the six companies had realized savings through their consolidation projects, IDC studied their IT expenditures over the last three years. It then compared these expenditures to the companies' estimates of what their IT expenditures would have been had they not consolidated to the iSeries. Besides looking for "hard savings" in IT expenditures, IDC also looked for "soft savings" that were realized through reduced downtime, increased IT staff productivity, and increased user productivity.
When IDC added up the figures, it concluded that the six companies had realized an average savings of $542,728 per 100 users over the last three years. As the table below indicates, this translated into an average annual savings of $183,716 per 100 users. Roughly half of these savings were hard savings that came from reduced spending on hardware, software, and IT payrolls. In most cases, the IT payroll savings were realized not from reductions in the number of IT staff, but from the avoidance of additional hiring to support standalone Intel servers. The remaining soft savings were realized from reduced downtime and increased productivity. Such savings are difficult to quantify and can vary widely between companies.
Three-Year Savings Analysis per 100 Users
Average Three-
Year Savings |
Average Savings
per Year |
% Share
of Savings |
|
Hard Savings (Cost Reductions)
|
|||
IT Payroll Savings
|
$124,428
|
$42,119
|
22.9%
|
Server Hardware/Software Savings
|
$145,787
|
$49,349
|
26.9%
|
Reduced Outsourcing Costs
|
$5,700
|
$1,929
|
1.0%
|
Soft Savings
|
|||
Increased IT Staff Productivity
|
$71,750
|
$24,288
|
13.2%
|
Increased User Productivity
|
$73,350
|
$24,829
|
13.5%
|
Reduced Downtime
|
$127,714
|
$41,201
|
22.4%
|
Total Savings per 100 Users
|
$542,728
|
$183,716
|
100.0%
|
[Source: International Data Corporation]
The IDC Report: Use with Caution
After reviewing IDC's report, I've concluded that, while it has some useful information for iSeries professionals, it contains several rather significant flaws. Indeed, I'm concerned that many Intel server sales teams--and possibly your own Windows server support staff--could point out this report's shortcomings rather easily.
Before I dive into those shortcomings, let me say up front that I'm in violent agreement with IDC on one thing: Companies that consolidate standalone Intel servers to the iSeries can realize substantial savings. That said, this report has inadequate information to let you determine how much you will save. Nor does it consider alternatives to an iSeries consolidation that could save just as much money. Among the flaws that I find in this report, the following are the most critical.
- An inadequate number of study subjects. Any student of statistics will tell you that a sample size of six companies has very little statistical validity. This study's validity is all the more suspect because IDC picked six companies from a list of firms that IBM provided. To make matters worse, IDC's assertions about Linux server consolidation are based on studies of just two companies. It is impossible to draw any conclusion about Linux server consolidation from such a sample.
- An inability to extrapolate the findings to smaller firms. IDC gathered its data from firms that are relatively large by iSeries standards. This is important, as it is easy to realize savings when consolidating large numbers of Intel servers that are running at low utilization rates and are located at multiple sites. It is far more difficult to realize savings at smaller firms with a single data center that has fewer than 10 Intel servers. The cost and savings models for such consolidation projects are very different from those presented by the companies in this study.
-
A refusal to consider whether current alternatives could yield similar savings. Clearly, the six companies in this study saved money because they consolidated Intel servers to the iSeries three years ago. However, if they were to undertake the same project today, could they save just as much money by consolidating older servers to multiprocessor Intel servers and then administering those servers using current systems software and best management practices?
While I don't have space to fully answer this question, I can tell you that today's partitioned, multiprocessor Intel servers can run multiple Windows and Linux applications at high rates of utilization and reliability. They take significantly less time to manage than their predecessors do, and their hardware costs are significantly lower than those for the iSeries. Don't get me wrong: The iSeries is still a great consolidation platform that can save companies plenty of money. However, today's Intel servers support consolidation in ways they never did three years ago. By not considering this fact, IDC's study leaves its readers with a conceptual blind spot. To put it another way, if you rely on this study to guide your thinking about server consolidation, you'll be driving your company into the future while looking in the rearview mirror.
In short, IDC's study points out something that we've known all along: Consolidating servers to the iSeries makes good economic sense. However, I would advise you to use the study with caution. Download it. Understand its strengths and weaknesses. Then, develop your own consolidation proposal that leverages the study's strengths, eliminates its weaknesses, and considers multiple options. It's the only way to create a strategy that can stand up under intelligent scrutiny.
Lee Kroon is a Senior Industry Analyst for Andrews Consulting Group, a firm that helps mid-sized companies manage business transformation through technology. You can reach him at
LATEST COMMENTS
MC Press Online