Research tools - laptops for the Internet and handhelds for mobile usability testing

isham research

Is the software roadmap for z/Series still z/OS - or Linux?

The new 40 MIPS z800 - the 2066-0E1 - ushers in a new strategy for IBM's mainframe range - the Integrated Facility for Linux (IFL) is standard. Despite its positioning as a new "General Purpose Model", the 0E1 is better viewed as a Linux model - like the highly unpopular 2066-0FL - with a slugged full-function engine attached for legacy toleration. This view is reinforced both by the legacy software price accomodation being limited to four years - enough time to migrate to Linux - and by the fact that an upgraded 2066-0E1 will always retain its IFL. As before, although the full-function processor is crippled to 40 MIPS by microcode, the IFL runs at full native speed.

The mandatory inclusion of an IFL engine in the 0E1 is no great surprise - manufacture of the z800 is largely outsourced to Hitachi and a 40 MIPS system priced pro rata with the z800-0A1 would sell at a loss. The 0E1 will thus be a somewhat more expensive proposition than a degraded single-engine full function system, but the emulation option remains for those who cannot justify the 0E1's mandatory IFL.

As noted elsewhere, IBM's greatest commercial success was with a mainframe using a public domain (free of charge and open source) operating system. Over recent years, mainframe growth has been stifled by high software prices - especially swingeingly high charges for small applications. VSE users have been especially badly hit by throughput losses caused by adoption of the Language Environment within a very strict Processor Group structure. Linux offers a way out, with the consequent shift back from software to hardware revenues naturally braked by the effort required to migrate.

The effect of the Damocles' sword hanging over a user who migrated to an 80 MIPS 2066-0A1 has been ameliorated but not eliminated - 40 MIPS is still an enormous jump for a small user. The proportion able to grow into and justify the software charges for such a machine is likely to be almost as small as for the 0A1, and the reversion to normal charges after four years may prove a great incentive to leave the platform, migrate to Linux or downsize to an emulation solution before then.

Unusually, one of the options announced was the ability to downgrade an existing 80 MIPS 2066-0A1 to a 40 MIPS 0E1. This option adds an IFL to the system, although this must be relinquished (or replaced with a new IFL) if the machine is upgraded again in the future. Adding the IFL costs very little, since the MCM is already fully populated - but it provides a way to increase hardware revenue to match the 80 MIPS variant. In essence, 40 MIPS' worth of $/MIPS is used to fund an IFL - IBM's revenue for an 0E1 is much the same as that for an 0A1, with the user swapping 40 MIPS of legacy capacity for 185 MIPS of Linux capacity - the maintenance price is actually slightly higher.

The largest problem most users will face - as with the 2066-0A1 - is ISV software charges. Although there have been some encouraging signs of realism at executive level, their sales farces continue to be rapacious. IBM contents itself with a suggestion that users should consider migrating to new IBM tools - a hollow laugh for the VSE community.

Comments on the above ideas are welcome at phil@isham-research.co.uk - all input will be treated in the strictest confidence - no contact of any kind will result unless requested.

Older articles are in the archive

isha research Home Page