IBM’s Math on User-Priced System i Versus Vintage Machines
October 29, 2007 Timothy Prickett Morgan
One of the phenomenal things about the AS/400, iSeries, and System i machines, and their predecessors, the System/36 and the System 38, is their electronic longevity. These boxes continue to do useful work long after the economic life of the systems is long since expended. OS/400 and i5/OS shops may be willing to pay a premium for such rugged machines, but they sure don’t like to do it as often as IBM would like. And so, Big Blue has to make the case for customers using vintage machines to move to new iron.
It is no different for Hewlett-Packard, which has the other remaining proprietary minicomputer platform, OpenVMS and which is in the middle of an upgrade cycle transition to a new version of OpenVMS running on Itanium-based Integrity servers. OpenVMS is celebrating its 30th anniversary this month, and Ann McQuaid, general manager of the OpenVMS platform inside HP’s Business Critical Systems division, says that there are over 300,000 VAXes, AlphaServers, and Integrity machines running OpenVMS and supporting over 10 million end user seats worldwide. But only after a year of having OpenVMS on Itanium in the field, with the launch of OpenVMS 8.3 last summer, are customers moving from testing and tire kicking to putting their largely homegrown application software in production on new OpenVMS and new iron. HP clearly wished that this had happened a lot faster, but when you are dealing with a conservative midrange base in the financial services, telecommunications, healthcare, and government sectors who can’t take downtime and that do not like messing with applications (which need to be recompiled in the case of OpenVMS running on Itanium), such transitions take time. And, by the way, while the technology inside the AS/400, iSeries, and System i servers makes it possible for applications to be upgraded to new iron transparently and automatically, the propensity to upgrade to a new System i box is still bound by the same simple economic argument that has kept many OpenVMS customers on vintage iron: This box works, and it is not a problem for me, like so many others. Why should I upgrade it?
IBM’s sales reps and those at its reseller partners must be blue in the face from hearing that so often.
Still, you can make a credible argument that the older iron has costs associated with it that could help pay for the newer iron, which provides more processing, memory, disk, and I/O capacity and opens up the possibility of consolidating other workloads now not on an OS/400 platform to one that runs i5/OS, Linux, and AIX.
IBM has done just that with a competitive analysis tool called the System i5 Economic Replacement Opportunity tool, which was used to generate a bunch of comparisons between vintage AS/400 and iSeries boxes and new user-priced System i 515 and 525 servers that IBM is showing off on the System section of its own Web site. The tool compares the average monthly hardware maintenance, monthly electrical power costs, and tier-based software maintenance costs of vintage AS/400 and iSeries models against new System i 515s and 525s. The presentation that IBM put together also shows the cost of the base server for the new System i 515 or 525 proposed as a replacement for the vintage iron. It also shows the monthly leasing costs for that server with a 36-month lease at a 5 percent lease rate with a 15 percent residual value at the end of the lease. The analysis also includes the amount of processing power (expressed in raw server and 5250 interactive capacities in CPW units) for the old and new machines. Note: These comparisons are not for push-pull upgrades, where IBM’s engineers come in and turn your old box into a new one. These comparisons assume that the old box is junk (which is many cases they are, but in some cases they are not) and that customers will dispose of the old gear through IBM or sell it to a second-hand dealer. Pricing is based on a three-year contract for maintenance and then averaged over 36 months.
Because you don’t have time to flip through the IBM Web site to see each of these comparisons individually that Big Blue’s marketing people have done, I have put them all together in a single table so you can see how the old and new gear stacks up. IBM’s sales reps will put together custom comparisons if you ask, and the System i5 Economic Replacement Opportunity tool is probably just a spreadsheet with data and formulas that allow any old AS/400 or iSeries box to be compared to new ones. But that’s a guess. At any rate, you can see that summary table by clicking here.
There were a bunch of errors in the IBM data, which I corrected in my table. IBM forgot to add in the power costs for old and new gear for a number of the machines, which I added back in. (Oddly enough, these helped make IBM’s case stronger, which is why I think someone was working too fast, as I often have to do and which drives me to make stupid errors like saying an F95 came out in 2003 instead of 1993, as I did in last week’s issue.)
As you can see from the chart below, when you add hardware maintenance costs, electricity costs, and software maintenance costs together for the old and new boxes and then subtract the new costs from the old ones, you can wind up with several thousands of dollars per month in savings.
Of course, those savings assume that customers are on a Software Maintenance contract and that they have their current hardware under maintenance as well. In many cases, this is not what customers are doing, and that is, of course, where IBM’s whole economic comparison falls right down on its face. But, for those customers who are on Software Maintenance and who do buy support contracts for their machines, the savings are so large that they can, in most cases, easily cover the cost on a lease for a new System i5 515 or 525 server. The proposed box swaps for the iSeries 800 and 810 machines do not cover the cost of the lease on the new iron. Which means customers should look for options other than the one provided by IBM in this presentation.
The push-pull upgrades are not just about getting current on new iron, but also about bringing a lot more computing power to bear on RPG and COBOL workloads and facilitating server consolidation from other platforms thanks to the sophisticated logical partitioning of the System i platform. Look at the increase in power that customers get if they take IBM’s advice:
The largest machine swapped out in IBM’s comparison had just under twice the raw CPW capacity, and it paid for itself through savings with a lot to spare. But as you can see from the chart above, in a lot of cases, IBM is saying that customers can increase their CPW capacity by a factor of five, 10, 100, or in some cases by more than 200 by moving to newer iron. That is a big jump for a conservative OS/400 shop to make, particularly when in a lot of cases you are asking a business owner to pony up what is essentially his or her own money to buy a lot more capacity than perhaps they need right this minute. But, that is of course always the challenge that server and operating system makers face as they are trying to get through their quarters and years.
That much is never going to change about the systems racket.