IBM claims its big blue tools penetrate markets
But has it stuffed a sock down its Power System pants?
SaaS data loss: The problem you didn’t know you had
It has become a habit of IBM to brag about how much market share its Power Systems servers are taking away from Hewlett-Packard, Oracle, and Fujitsu as it closes out each quarter, and the fourth quarter was no exception for Big Blue.
The Power Systems lineup includes older servers based on dual-core Power6 and Power6+ processors as well as four-, six-, and eight-core Power7 processors; these machines run IBM's AIX Unix variant, its proprietary IBM i (yes, that is a stupid name for an operating system) and OS/400, and Linux variants from Red Hat and Novell.
The Power Systems lineup was refreshed in stages throughout 2010, starting with blades and midrange rack and tower boxes in February, enterprise-class bigger boxes in April, and entry rack and tower servers and the Power 795 behemoth in August.
As El Reg previously reported, in the fourth quarter, IBM's sales rose 6.6 per cent, to just over $29bn, and net income rose by 9.2 per cent, to $5.26bn. But Power Systems sales were only up two per cent (six per cent if you count the systems software such as hypervisors, operating systems, clustering, and other tools).
The new entry Power Systems machines did better against an easy compare, rising 30 per cent, and midrange boxes saw a seven per cent bump. Which means big Power-based boxes are still not selling as well as they need to.
IBM's mainframes, which were refreshed in July last year, had an impressive 69 per cent revenue spike, by comparison, and System x and BladeCenter sales rose by 18 per cent, with sales of big x64 iron rising 30 per cent in the quarter.
If IBM is having trouble at the high-end of the Unix and proprietary systems market, it is a fair guess that its competitors are not faring a lot better. (You have to guess because HP and Oracle don't talk about their respective Itanium and Sparc products in any detail.)
Selling into the existing customer base might have been tough for the Power 770, 780, and 795 machines in the fourth quarter, but IBM continued to get some traction peddling Power boxes and related software and services into non-IBM accounts.
According to Mark Loughridge, IBM's chief financial officer, the company did 280 competitive displacements with Power Systems in the fourth quarter, driving around $325m in revenues. This, Loughridge said in speaking to Wall Street after Big Blue reported its numbers, was the biggest quarter ever for competitive replacements for Power iron. He added that IBM has done over 1,000 such replacements in 2010 and generated nearly $1bn in revenues from these.
IBM never says how much of the competitive takeout money is from hardware, but it is probably about 25 per cent for systems, including memory and local storage, plus another 25 per cent for storage. The rest would be for software and services. These takeouts are not just important for the hardware sales they generate in the quarter, but the future hardware sales they anchor in coming quarters and years.
They also take money away from competitors: in 2010, Loughridge said about 60 per cent of the Power Systems competitive wins came at the expense of Oracle, and 30 percent came from wins at HP accounts. Some deals are from consolidating workloads that move from various x64-based servers onto Power, and IBM did 100 such deals in 2010.
It would be interesting to know how many accounts IBM lost throughout 2010, quarter by quarter, to see if Big Blue is gaining ground or losing it. For instance, it seems very likely that IBM lost more than 1,000 accounts that formerly used its AS/400 and iSeries minis, which are based on the same Power platform as the AIX boxes.
In 1998, IBM had over 275,000 AS/400 accounts, but said last year that it has somewhere north of 100,000 IBM i and i5/OS accounts (those being the most recent names of the venerable OS/400 operating system). If you do the math, over the past five years that means losing 25,000 OS/400-i5/OS-i shops each year as IBM adds around 2,500 new accounts per year. That is starting from a base of around 220,000 in 2005, a number that IBM itself supplied to me at the time.
It looks to me like IBM is taking good care of its AIX base, but has not figured out how to make its IBM i base happy. Unless it is peddling System x boxes to them, of course.
How many mainframe accounts did IBM lose in Q4? In all of 2010? How many accounts did Oracle, HP, and Fujitsu take away? How much of the Power Systems revenue stream is dependent on Oracle databases? Oracle and SAP applications? So many questions.
Another thing I would love to know is just how much of IBM's hardware sales in the fourth quarter came from IBM's own Global Services group. As one of the largest server buyers in the world and one of the largest data center operators, with zillions of outsourcing customers who host things on mainframes, Power Systems, and x64-based machines, Global Services can in theory help IBM engineer a good revenue figure for any of its product lines. Just like IBM can engineer its earnings per share growth by buying its stock back with cash.
For all I know, IBM Global Services customers get iron that is one or two generations back that Big Blue has sitting around in the warehouse. But those count as revenue, too. And they can sell at list price or whatever figure Big Blue sets, too - which is why I think IBM should be compelled to tell Wall Street how much of its own wares it buys for its own IT operations, as well as for supporting Global Services customers.
It will be interesting to see how the System z mainframe does in the four quarters of 2011. Will the spike in Q4 2010 sales be an anomaly, a function of pent up demand - or does it signify the beginning of the long tail of a mainframe sales revival that will continue for the next three years? ®
COMMENTS
275K --> 100K for IBM i / i5/OS / OS/400
It's a shame that so many organisations appear to be/have been moving away from IBM's AS/400 platform (in all of its various flavours).
The AS/400 is, by-and-large, the best box that IBM ever designed, its OS having been developed by a team specifically chosen for their "lack of exposure" to UNIX and other POSIX-like operating systems.
I would venture to say that in our relatively short history of computerdom, there have been few other boxes that "just work" after being removed from the shipping crate, moved into the server room, plugged in, and powered on.
Long live the 400!
AS/400 Cash Cow
The AS/400 was a good box, but at what price? IBM knew it had no natural competition once it was installed in an account so IBM milked customers for every penny! And because it didn't lend itself well to external disk, IBM were able to charge the earth for disk as well! The exact same 'bits' for a PSeries or ISeries box were priced very differently.
Small box focus?
Back in Feb 2010 I did ask our IBM salesgrunts why were IBM starting with the Pee7 entry systems rather than the big-daddy 795 (OK, I teased them by implying it was because they had to get something out before Nehalem took all the sales, and was it becasue it was easier to get the small boxes out the door first?). Their neat answer was that their primary targets for the year were legacy Solaris accounts feeling left in the lurch by the Sunset, and they didn't need the 795 to target the majority of those accounts. It doesn't seem to have been such a bad idea going on the sales figures. Maybe their attention will shift more to the top-end once they've finished feasting on the Sun carcass.
IT infrastructure monitoring strategies
Agentless Backup is Not a Myth
Top 10 SIEM implementer’s checklist
Steps to Take Before Choosing a Business Continuity Partner
Enabling efficient data center monitoring
