IDC Quantifies the iSeries Payback for Server Consolidation
by Timothy Prickett Morgan
One of the things that the most recent top executives in charge of the iSeries Division promised to do when they came into power last year was get the word out that with analyst reports concerning the technical and economic advantages of the iSeries platform. A few weeks ago, these IBM executives were talking up the fact that there are now 19 analyst reports out there in the field concerning the iSeries. (I sure have not been able to locate that many, but the Web is a big place.) One of them comes from IDC, which IBM usually turns to in order to make the economic case for the iSeries.
As many of you know, I prefer for the iSeries platform to take a head-on approach when it comes to competing against Unix, Windows, and now Linux in the data center. I want market studies that show how the iSeries competes against these platforms in small, medium, and large configurations doing real data processing work, since this is, by and large, what the AS/400 and iSeries platform has been charged with doing at the 200,000-plus customer accounts where it is humming away today. It has been a long, long time since anyone has done such a comparison, and there are reasons for that. First, IBM is not trying to price against Windows, which dominates data processing at small-sized, and increasingly medium-sized, businesses, but rather is using Unix platforms (which are, in many cases, overpriced) as the benchmark against which it measures the iSeries. This is all well and good, but the kinds of customers that we want to buy iSeries machines--small and medium businesses looking to install ERP applications--don't tend to want Unix boxes and they tend to think of any computing as Windows computing. They could care less about Unix or OS/400, and they have no idea what i5/OS is. This is one of the reasons why I have been championing the idea, time and time again, year in and year out, of creating a low-cost, turnkey iSeries system that runs ERP applications suitable for SMB customers. Such boxes would have to compete head-to-head with Windows, including the cost of the ERP software against Microsoft's own various ERP suites. If you want to sell a lot of iSeries boxes to new customers or those who are frustrated with Windows--and if my experience with Windows on the desktop and on the server is any indication of the frustration level, it is very, very high--then this is precisely the way to do it.
But, alas, Big Blue doesn't see it this way. So we don't have such products and we can't have such analyst reports that compare these mythical products--I called them puppy AS/400s, puppy iSeries, the iDeal iSeries, and now the System i5 for SMB over the years--to actual Wintel platforms running ERP software.
Still, all is not lost. When you are defending the iSeries platform, which all of us do every day, every arrow in your quiver is something to keep the Unix snobs and Wintel mobs at bay. And in this regard, the IDC report that IBM commissioned analysts Jean Bozman and Randy Perry to do is useful.
The report has the unwieldy title of Infrastructure Simplification and the ROI of Consolidating Windows, Linux, and AIX 5L Applications on IBM eServer iSeries Systems, and just in case you didn't know, "infrastructure simplification" is one of the things that Mark Shearer, the current iSeries general manager, cooked up for IBM's Systems Group a few years back as a new twist on the old server consolidation marketing push. There's nothing wrong with infrastructure simplification--it's a great idea, and something that most IT shops (including my own) radically need. The idea is to use the virtualization technologies to reduce the number of operating system types as well as the physical number of servers in the data center, which leads to higher server utilization and lower administration costs, both of which turn into cash savings over the long haul. This is precisely the argument that IDC is making in its latest report for the iSeries.
In the past, IDC reports have talked about the cost of the iSeries platform compared with Windows and/or Unix boxes running ERP workloads, but rather than finding customers who use the exact same workloads, they just treated all applications the same and found that OS/400 shops spent less money than Windows or Linux shops. I crabbed quite a bit about this, since it is not exactly fair to compare an OS/400 shop using homegrown RPG applications that change very little against a Unix shop running a Web-enabled suite of software from one of the giant vendors. No offense, but OS/400 shops don't go in for crazy expensive software that hogs resources, so of course they don't spend a lot of dough to support their applications. In many cases, OS/400 applications in real data centers have been refined over decades, and they work like a top. But they ain't exactly portable across platforms, and that is a cost just as much as it is a benefit, and you need look no further than iSeries pricing compared to alternatives to know that I am right.
The fact is, any new iSeries customer is almost certainly going to buy third-party application software, not code their own in RPG or COBOL or (heaven help them) Java; and any existing OS/400 shop looking to modernize their apps by throwing them out and buying applications off the shelf is going to do the same. (There are very few of either, and most shops will modernize their applications at their own pace, using a wide variety of tools and techniques.) So, in my opinion, these past IDC comparisons were of limited value outside of the obvious anti-Windows and anti-Unix propaganda value they have for iSeries shops trying to defend the platform. These studies didn't help sell new iSeries boxes so much as comfort those who had them.
What the IDC analysts did this time around to derive their cost and return on investment metrics is interview a dozen iSeries shops in the United States, Europe, and Asia who have consolidated various OS/400, Linux, Unix, and Windows workloads in their data centers onto the iSeries platform. In some cases, customers were consolidating multiple AS/400s onto a single iSeries as well as putting Windows workloads on the Integrated xSeries Server; some customers moved Linux and AIX workloads onto the iSeries. After running the numbers for the 12 different companies through a magic ROI calculator that IDC created in conjunction with Alinean, a company that creates return-on-investment calculators for IT suppliers, IDC reckoned the average value of the iSeries as a consolidation platform. These results were then normalized for 100 users at each company.
Specifically, IDC and Alinean found that the three-year ROI for consolidating workloads onto the iSeries was 258 percent. That ROI was calculated in the following fashion. IDC figured out how much money these companies didn't spend on extra administrators and extra server iron, and then added to that the value of the increased productivity of both the IT shop (which had less administration to do) and end users (who had servers that were more available). On top of this, the value of not having so much downtime was plugged into the equation. The cost reduction in IT administration and server spending represented 58.9 percent of the savings, with IT productivity being 6.5 percent, end user productivity being, 13.6 percent, and increased uptime (the lack of downtime, which is a cost) being 21 percent of the savings.
The initial savings in the first year came to $67,017 per 100 users, with a total of $147,194 per 100 users over three years, yielding an average of $51,626 per 100 users in cost savings. The upfront investments for these iSeries shops in the first year came to $26,282 per 100 users for the first year, with a total of $35,139 per 100 users for three years. That averages out to $11,713 per 100 users per year of cost for three years. The ROI calculation is a bit complex. You take the $147,194 per 100 user savings over three years and discount it by 12 percent per year to take into account the time value of money. Then, you take out the total investment (also presumably adjusted by the time value of money) and you get a net present value of savings worth $90,553 per 100 users. You divide this by the total investment ($35,139 per 100 users) and, voila!, you get an ROI of 258 percent. Then, you calculate the payback period by taking the investment and dividing it by the first year cash flow generated by the consolidation, and you see that the iSeries consolidation pays for itself in the first 5.5 months.
Simple, right? (If you read the report, which you can do at the link at the bottom of this story, you will see the math is a bit hairier than this and that IDC has not really provided the details behind these average numbers--but it has promised to do so in a much fuller report due sometime in the future.) I can just see the prospective IBM salespeople or business partners trying to explain this math to an iSeries shop who probably is already predisposed to do such a consolidation. And, as I have said, this report doesn't do much to convince non-iSeries shops to consider the platform.
Infrastructure Simplification and the ROI of Consolidating Windows, Linux, and AIX 5L Applications on IBM eServer iSeries Systems