The Way IBM Sees New Versus Prior i Platforms
May 27, 2008 Timothy Prickett Morgan
In the wake of any major server launch, IBM‘s top marketeers have usually put together two sets of comparisons for IBM’s own sales reps and its business partner resellers. One set of comparisons looks at the new product line and compares and contrasts it with the immediately prior product line, usually highlighting scalability increases or price/performance improvements for given configurations. In the AS/400 family, IBM used to make a lot more comparisons to Unix and other proprietary platforms, too, which helped customers remain confident that the platform was competitive.
The latter type of competitive platform comparisons are not usually available to the public, and are especially thin on the ground in recent years as the iSeries and System i platform did not keep pace with the insane price/performance improvements available in the Windows servers that now dominate the small and medium business arena. IBM has always been very good about giving lots of relative performance metrics within the OS/400 and i5/OS product line, and in recent years has been happy to even share priced configuration comparisons–real-world comparisons–as it moved from server generation to generation. This has all been helpful for OS/400 and i5/OS shops who need to make the technical and economic arguments for upgrading to newer platforms.
But this time around, with the Power Systems family of launches, the data seems to be sparse. While the IBMers I have spoken to have said I should take a look at the aggressive pricing for maintenance on the new machines versus earlier System i boxes–which I will do in the coming weeks–it has been difficult to get the kind of platform comparisons I am used to getting my hands on. This, I believe, is a function of executives moving from the former System i division, where they did their homework, to the Power Systems division, where the AIX folks maybe did their homework in comparing old and new AIX platforms or maybe they didn’t. In the Unix world, where there is actual competition, you don’t worry about comparisons to last year’s model of your own gear as you do about how you look next to this year’s model coming out of Hewlett-Packard or Sun Microsystems.
As I explained in last week’s issue, by IBM’s own reckoning, the vast majority of the i platform sales in 2008 will be for entry machines in the line, which include last year’s user-priced i5 515 and 525 boxes and this year’s Power 520 M15 and M25 boxes. And according to the documents I have seen, IBM has not priced the entry configurations much differently between Power5+ and Power6 systems, and the performance is not radically different, either. This, I think, is absolutely intentional.
I have waded through a number of different presentations that IBM put together for sales reps and business partners, and over and over again it says the same thing. First, that by moving to the System p pricing structure for hardware, what would have been called the System i (but is now a Power Systems box running i 6.1) has more competitive hardware pricing. That means the new machinery is competitive with other Unix and OpenVMS gear, presumably, which is what IBM is pricing against because i for Business is a serious business platform. (The fact that most SMB shops don’t know what i5/OS, Unix, OpenVMS, or Linux is seems to be beside the point. . . .) The other phrasing you hear, again and again, and this has to be something that the System i reseller channel must have been pushing for, is to “maintain the total solution price.” In plain English, that means the difference in price and price/performance between the System i 515 and 525 and the equivalent Power Systems boxes just launched will be minimal. Take a look for yourself at IBM’s own presentation.
What IBM calls this, as you can see, is reflecting the value of the System i in the i software stack instead of in the hardware, where I have argued (since 1993, I think) that IBM should not have put it in the first place. (Big Blue had other needs, such as artificially inflating hardware revenues in the 1990s and early 2000s, but I think this practice really hurt the competitive stance of the AS/400 and its follow-ons.) As you can see from the chart above, there is not a big difference between the Power 520 and its System i 515 predecessor; the base M25 has a lot less expensive hardware price than the 525, by IBM’s reckoning, but the software bill makes the newer box only slight less costly.
IBM has good reason to try to keep the prices all the same, despite the competitive pressures from Windows boxes that now have four cores per socket. It has to give the reseller channel enough margin to invest in the relatively hard sell of pushing Power6 servers running either i5/OS V5R4 or i 6.1 rather than just give up and push Windows and a migration strategy away from the i platform. Whether or not this is a smart long-term strategy remains to be seen, but this is the play that IBM’s marketeers are calling.
The other differentiation that IBM is making between the System i 515 and 525 and the new Power 520 is that the older boxes support i5/OS V5R4, which many customers can more easily certify their applications on since i 6.1 requires program conversion. (See i5/OS V6R1: The TIMI, It Is A-Changing and i5/OS V6R1 Compatibility an Issue for Software Vendors for more on that.) Moreover, customers who need to support legacy twinax devices and quarter-inch tape drives have to go with a 515 or 525, since these are not supported on the new Power 520. The other thing IBM is stressing in its presentations is that maintenance is lower on the new boxes than on the older ones, but just how much different, it has not said. (Presumably someone has done that math, but I haven’t seen it yet so I am working to do it myself.)
Perhaps more significant is the suggestion by IBM that for many customers on Power5+ System i 520 boxes, it will be less expensive to move to the user-priced i 525 than to go with a Power6-based M25 machine. The main reason is that the upgrade from the old 520 to the 525 allows for the reuse of many parts within the system (memory cards, peripheral cards, disk drives, and such), while the jump to the M25 means replacing a lot of these components.
This doesn’t sound like a very aggressive plan to push Power6 iron, and I think this is no accident. I can’t prove anything, but the pricing and packaging of the i variants of the entry Power Systems machines leads me to believe that IBM still is not getting good yields on the chips. IBM is just as happy to peddle Power5+ machines in 2008, and I would guess for good reason. It wants every Power6 chip it can make to end up in an AIX box, a blade server, or a high-end AIX or i enterprise-class server. I would have expected IBM to compete more aggressively on price/performance if chip yields were higher, and the fact that IBM is offering customers a 10 percent performance bump (if they don’t end up being memory or disk constrained in their configurations) at essentially the same price is a telling fact.
In any event, I will be ginning up more precise comparisons than the ones I have seen from IBM to give you a sense of where things actually stand over the course of the past several i-related products.
One final note. Brian Podrow, a long-time AS/400 manager within IBM and lately a Power Systems product manager for the Americas group, has retired from Big Blue after 34 years, effective April 30. Like many IBMers, Podrow started out with the System/3 line in Rochester, and has been one of the key thinkers behind the positioning of the AS/400, iSeries, and System i lines. In the past decade, since I have known of Podrow, he has worked behind the scenes to support resellers and IBM’s direct sales force with the Quick Pricers and competitive analysis documents that are a vital part of sales and marketing of the platform. While I was never officially supposed to have such documents, I can say from personal experience that Podrow’s work has helped clarify IBM’s positioning to me in a detailed manner that you just do not get anywhere else inside of IBM or among any other IT suppliers, for that matter. Podrow and I did not always agree on what some data point in the product line might mean, or what IBM could or should do next to be more competitive, but his work on behalf of the AS/400 and its progeny was nothing short of prodigious, and vital, too. Brian, I hope that means something coming from a blabbermouth like me. I mean it sincerely, and I wish you all the best as you take your well-deserved vacation and pursue the business opportunities you have put on hold for 35 years. Thanks, from all of us. Your work is already missed.