Technology News Technology News

Another Perspective


IBM's most economical servers, the System x machines, have been booming for a year, bringing Big Blue a revenue boost of more than 30 percent.  At the other end of IBM's product range, multi-million dollar mainframes have enjoyed a 15 percent jump in sales.  But in the center IBM is in trouble.  Big Blue's best efforts have failed to produce a turnaround.  For the past six quarters, Power Systems server revenue has been down at least 10 percent and sometimes much more.  You can't tell by the company's confident posture, but IBM is in crisis.

revenue by line
Line By Line
IBM is a year into the x bounce, one quarter into the z bounce
and still in the doldrums

Maybe IBM's troubles are just a blip.  Maybe the new Power7 boxes that IBM says have more appeal than their predecessors will finally yield a turnaround, the one that should have become visible during the third quarter.  But then again maybe not.  Maybe there's something really wrong with IBM's midrange sales effort, and maybe it's more than just the midrange, too, because these days Power Systems overlap X64 servers at the low end and match or possibly exceed IBM mainframes at the very top.

If you step back and look at IBM's total server revenue, the picture isn't quite as ugly.  Booming sales of X64 equipment and a solid gain in mainframes brought IBM an 8 percent increase in total server sales.  It was the best news in a while.  IBM's server revenue was hit very hard in 2009 and, despite a jump in System x sales during the first quarter, nothing else in the IBM processor line was doing that well until after the middle of this year.

But now it's the fourth quarter and the battle to sell servers for new applications (and, sometimes, to re-host established applications) is a free-for-all.  Intel and Advanced Micro Devices have excellent server chips.  Compared to the CPUs available last year, they are fast, they are cheap, and they are green.  Worse, for IBM's Power peddlers, both Intel and AMD keep rolling out versions of their current motors that reflect ongoing efforts to refine their chip designs and fabrication processes.  Moreover, now that demand is on the way up, both Intel and AMD can afford to replace their circuit designs very quickly.  This potential for change is an advantage that is unavailable to IBM with its Power or mainframe chips or the Sparc producers, Oracle and Fujitsu.  It is not even an option for Intel's Itanium line.

If IBM wants to build up sales of its Power System line, it has to pay a lot more attention to its sales effort.  It has to dramatically improve the perceived value of its Power-based servers.  It also has to do a lot more to make the mainframe more than just a high tech hospice for legacy applications, because the image of the mainframe influences the image of everything else that carries the IBM logo.

At the heart of IBM's sales problem is a legend that apparently still thrives within IBM, a myth that IBM is usually seen as a company that knows much more about computing than its customers.  And because it is seen as so wise, IBM imagines, it can guide its users toward the products it wants to sell.  But this isn't actually the case, and it cannot be the case when everyone knows that prominent new companies that have turned computing in to gold – Google, eBay, Amazon, and Facebook – and established outfits that have been able to redefine their customer service strategies are very likely to be using X64 hardware, Linux environments, and, heresy that it may be among users of big and pretty big iron, Windows, too.

server revenue
Server Sales
IBM server revenue is moving in the right direction,
but too slowly

There are lots of reasons for this, but I can't think of any that is beyond remedy if IBM starts using its brains again, although it's been a while since I have seen any clear signs of clever and creative selling coming out of Armonk.

IBM seems to think it can get so much business at the country club and on the golf course that it doesn't really have to go out and sell things the way Apple sells iStuff or Hewlett-Packard peddles servers one person can lift.  Eight-figure services deals are undoubtedly cut that way, and IBM remains a leader in that aspect of the computer business.  But it's doubtful that IBM can preserve its Power server business if it seems disinterested or aloof, even if its sales failures are really the result of sincere incompetence, not indifference.

I don't think IBM has to do all that much to get its Power products to sell better.  They are very fast computers and they have pretty rich and reliable operating systems, databases, and other software.  And while there's a pretty big gap between the boxes and systems software IBM sells and the applications layer end users require to do their work, IBM's ineptitude has not killed off the ISV ecosystem that makes the IBM i operating system so worthwhile and, for some purposes, makes AIX a viable platform for high-end ERP setups.  \r\nNevertheless, the whole atmosphere in the X64 market is much more lively than that in the Power world, because the X64 market is really competitive.  There are at last three and often more than three major vendors for every size server.  Each of the vendors has a zillion resellers who will happily cut each others' throats to land a sale.  And that's before you stir in software and applications support outfits.  On a good day the X64 world is like Jurassic Park and the rest of the time it's really fierce.

By contrast, IBM has its resellers worlds (and not only the one with Power in it, the mainframe world, too) performing like a North Korean military parade.  There's no way an end user can bang resellers' heads together to get a better deal or a little more service.  It may well be that every Power deal is a fair one and that every reseller is deeply committed to customer satisfaction but, frankly, how can an end user tell? It's easier to change garbage collectors in the Bronx than it is to change a Power Systems reseller, and that's a soft world compared to what goes on in mainframe country.

If IBM Power boxes and mainframes were reliable but X64 hardware was not, that would offset the allure of a competitive market.  But that's not the case.  Servers and storage in the X64 world work very well and users can get pretty much all the redundancy they want.  If there really is a difference in uptime, it's a couple digits past where anyone would care because by that point the thing that takes a system down is probably software, maybe some the user's own people wrote or tinkered with.

It's not that bigger iron isn't better.  It is.  But it's not enough better to compensate for the extra bang for the buck that's part of the X64 world, nor can it provide the terrific software choices that are available to anyone developing Internet applications in either of the two most popular environments, Linux and Windows.  It's not that IBM isn't trying.  It wasn't long ago that Big Blue made a nice effort to promote PHP for i5/OS and then IBM i and to stand behind some plumbing that would let PHP technology that was developed to work with MySQL hook up to DB2.  But that's such a kludge.  Users can't be happy paying a premium for IBM's proprietary technology and then living on the planet catch-up.

IBM has to come up with some kind of sales proposition that makes Power Systems running IBM i or AIX systems a lot more attractive than they are right now.  If it doesn't know quite how to do this, it might try talking to its customers.  In fact, it might try talking to two (overlapping) groups of customers.  One group is the Power base.  What do these people want? What would make them invest more in their Power Systems? As for the other group, that's the System x base.  IBM ought to ask them why they are willing to pony up 30 percent more this year than last.  Maybe they would help IBM develop the elusive feature the System x has but Power currently lacks: motivation.

— Hesh Wiener November 2010

Copyright © 1990-2019 Technology News of America Co Inc.  All rights reserved.