Sometimes, the server business reminds me of driving in a car on a long trip with a lot of kids in the backseat. You just want to reach back and start slapping and tell everyone to shut up and stop talking nonsense. But fathers today are more civilized (at least outwardly) and don't do the things their fathers did (judge for yourself if civilization is better for it). And so it is with the tit-for-tat noise coming out of IBM and Hewlett-Packard about who is replacing whose iron at data centers around the world.
HP started the latest round of squabbling about server replacements - which is always the rage when the economy gets tight, by the way - when it declared two weeks ago that its missionaries had converted more than 250 mainframe shops to its Itanium-based Integrity servers in the past two years. Most of those shops have moved to HP-UX, but some have moved to Windows or the NonStop fault-tolerant operating system, and many also deploy Linux on Integrity boxes beside the primary operating system they deploy applications upon.
If you remember in that story, I added a key point: There are only about 10,000 mainframe footprints in the world. So as Monty Python might remind us: Every 'frame is sacred, every 'frame is great. If a 'frame is wasted, Sam gets quite irate.
So, Big Blue cranked up its PR machine, and pulled some statistics out of its, er, sales database and wanted everyone to know that more than 5,000 companies worldwide have replaced iron from HP, Sun Microsystems, and EMC since 2004 and moved to them to IBM alternatives.
Breaking this down a little, IBM is claiming that in less than one year, more than 150 customers have moved onto its System z mainframes from HP and Sun platforms. When pressed for more details about where these customers were coming from, IBM's PR people said they would get me some answers, but all I got was static. And since the inception of the so-called Migration Factory that IBM set up "several years ago," more than 1,300 customers have been moved to Power-based servers from Sun and HP platforms, and this year alone, another 800 customers have moved onto System x iron (which presumably also includes BladeCenter blade servers).
Yes, IBM is mixing different categories and different time scales, but this is what happens when people bicker. On the storage front, IBM says that since 2006, it has moved over 2,900 customers from EMC iron. (Any minute now, EMC will pipe up with how many IBM, HP, and Sun takeouts it has garnered).
HP leaves out the 'sponsored' bit
In making its announcement, IT consultancy Robert Francis Group managed to get scratched by HP's initial press release, and now it had protested. Here's what HP's original press release said: "According to Robert Frances Group, a leading provider of consulting and research, the capital savings range from $1.5 million up to $23 million, with up to an additional operational cost savings of more than $4 million over four years. The study shows HP Integrity servers consumed 41 percent less energy and used 48 percent less space than the IBM z9 mainframes they replaced."
If you look at the study that HP cited, which is posted on Microsoft's site and which came out in July 2008, the comparison inside this report clearly shows an SAP setup on a mainframe being a lot more expensive than a setup on an HP Integrity server running Windows Sever 2003 - and mostly because mainframe software and maintenance are very expensive, not because of hardware or energy costs.
"In some cases, maintaining legacy systems on mainframe equipment makes the most sense," the RFG report concludes. "However, RFG finds that there are many times when the combination of open systems, based on HP Integrity servers and Microsoft software, delivers the optimal combination of low acquisition costs, low administrative costs, and an efficient use of electrical power that make a compelling TCO argument."
But, alas, that is apparently not the conclusion that we are meant to draw from the report - at least not based on the statement that RFG put out last week. And from that statement, we learn that HP apparently sponsored the report, which somehow still has Microsoft copyrights on it. (Go figure).
"In a recent report sponsored by Hewlett Packard Co. (HP), RFG discussed the advantages of upgrading old computer systems to new energy-efficient ones. Last week HP issued a press release that referenced this report in which readers can infer that RFG believes a new HP system is always less expensive than a new comparable IBM System z. The report did not make such an analysis and made no comparative statements of that nature. This is a misleading conclusion that RFG does not support. RFG has long stated that the mainframe is one of the best and most energy efficient platform options. RFG has written many research reports stating that mainframes should be considered and used in certain environments and RFG stands by those statements."
If RFG really wanted to have anyone reading this report draw that conclusion, then maybe a report that wasn't sponsored by anyone and that showed the scenarios where keeping a mainframe made sense as well as those that showed where it didn't might have helped. Which is why sponsored research is always suspect, as are vendor statistics, now that I come to think of it.
But we have to make do with the "information" we have, right? ®