isham research
Major component (detailed analysis once the announcement has been digested):
"Capacity Upgrade On Demand"™
See Amdahl's "Dynamic QuickCapacity" feature on the Millennium 700 Server. IBM's facility has other components such as "Concurrent Conditioning"™, "Planning Ahead"™ & "Processor Unit Optimisation"™. IBM does not appear to have an equivalent to Amdahl's QuickMemory feature. A number of new instructions will be provided. Chief among them is "Store System ID", a much expanded corollary to Store CPUID adding a little more support for the above features. This instruction is being disclosed to ISVs in 1Q1999. The ones isham research have spoken with are unimpressed, and it looks like being taken up by as many as have taken up other IBM pricing initiatives over the past few years. Not many.
IBM's most lucrative current market is the sale of the installed but unused processor chips in G5 systems. If these systems are swapped out without being first 'upgraded' IBM's effective cost of product for G5 will increase significantly.
The anticipated 'Y2000 lockdown' frightens IBM. If they can persuade the user base to install lots of disabled G5 chips and convince them that massive effort has gone into simplifying 'switch on' upgrades, they may be able to create a potential market within those organisations that have banned physical system swaps during 2H1999.
For this to be successful, users need to be persuaded that it will work, and that G6 will not be an inevitable upgrade for most of them. This implies freezing functionality between G5 and G6 (perhaps a Statement of Direction) and talking down G6 performance - hence the '+25%, not +35%' story that I believe has been placed deliberately. This story is inconsistent with the 'three-fold scalability by 2001' presentation by Bob Stephenson, and with the performance non-disclosed for parallel technologies in RS/6000 and AS/400.