IBM's growth continues to be the envy of other hardware vendors - Mainframe Markets

Rethink IT, March, 2004 by Berry Graham

MARKET OVERVIEW

The year ended very well for IBM, with financial figures that are the envy of all other hardware based vendors and even stone purely software and services-based vendors.

zSeries sales in particular were very good and represent yet another year of growth from the platform--a continued curve that has at no point in the recent passed stalled. Other vendors are beginning to get back into growth, but from a level that is far lower than a few years ago. However, IBM is not complacent and has a large number of new mainframe announcements planned for this year.

These will almost certainly all be made around the middle of April, which will see the fortieth anniversary of the 360 Series mainframe announcement. This was an announcement that forever changed the face of computing with a solid base, suitable for all types of users (the 360 name represented 360 degrees) that could evolve effectively 'forever'. For users of Sun, HP, or Microsoft-based solutions this must seem unbelievable as it effectively means that applications written 40 years ago can still be running today with no changes--indeed the instruction set itself has only evolved since 1963 to allow for new instructions with (most of) the original ones still available today.

The centerpiece of the hardware announcements will undoubtedly be the z800 replacement (named here as the z880, but don't be surprised if IBM uses a different name). A turbocharged z990 is also just possible, although it may be a bit early for this system.

For users of older CMOS technology such as the G5/G6 who perhaps see the T-Rex as too big, but the z900 as out of date, this new system will be the perfect solution and they should therefore await the z880, which will certainly ship in the first half of 2004 although probably not on the date of announcement. As previously reported, this new system will provide at least twice the capacity of the largest z800 available today and use z990-type technology throughout.

Technological advantages for the z880 and z990 are likely to include a Java engine that effectively runs Java code without requiring any of the S390 processor capacity. However, as with CBU (capacity back-up), IFL and ICF engines, the Java engine (or engines) will take up valuable processor slots on the systems. This will ultimately push IBM into replacing the z990 with the faster and more engined follow on system by mid-2005 at the latest.

IBM will also continue to deliver lower software costs for the mainframe platform through a combination of some extensions to the low end software offerings and by offering the T-Rex 10% reduction in MSU's for software pricing on the z880 as well.

This means that a marketing ploy from IBM to compensate for a performance deficiency in the z990 (the 10% MSU reduction for software pricing) will become a standard way of reducing software costs for IBM users into the future. It's just a shame that it confuses everyone when comparing hardware and software MSUs and this can only get worse in the future if a user is trying to review the past growth. Just what was the capacity in 2004--2,000 or 1,800 MSUs? Was that the hardware or software figure? It's bad enough now without these extra complications.

Interestingly, IBM is trying hard to convince me that the z990 model 10% reduction in software MSUs is not in any way connected to a performance deficiency, but is instead aimed at giving IBM mainframes a software advantage against its competitors. In fact, I believe that there is some truth in this and that the 10% software MSU reduction is intended to improve the platform TCO, but as yet I am not convinced that the performance deficiency does not exist.

The IBM direction on platform costs is to reduce the mainframe software costs by around 20-25% per annum, such that a user can grow his capacity at this level without (significantly) increasing his software costs. Obviously this is based upon the current software portfolio and hopefully all users realise that IBM intends to increase the amount of IBM software in every account to grow their software revenues in this way. In addition, it intends to grow the platform by winning users from other platforms and making the zSeries the platform of choice for new e-business applications and computing on demand.

However, users who do not grow at 20-25% per annum are unlikely to see any significant reduction in their software costs as IBM will cleverly manipulate software pricing to ensure that only growing users can really reduce their costs--and remember this is only a cost per Mips reduction, not a real cost reduction in the software bill.

Thinking about the TCO of the platform, which readers are aware my studies show is under half the cost of any alternative already, the following advert from Oracle in the Financial Times at the end of 2003 made interesting reading.

the Oracle grid turn 64 small servers into a giant mainframe

it's cheap ... and it never breaks

ORACLE

oracle.com/grid

Note: 'never breaks' indicates that when a server goes down, your system keeps on running.