Word from across the Atlantic is that a key reason for the hold-up in launching the new RTs was specifically the applications gap, but the hold-up does nothing to enhance the credibility of the machine, because IBM designers that have left the company say that they were asked to design to a performance specification and a price, and grew terminally frustrated when the company failed to launch their brainchildren until 18 months after the target date, thereby missing its window of opportunity: IBM has always been able to get away with that with its proprietary systems because however late the AS/400 may have been for despairing System/38 users, they had nowhere else to go, while in large-scale mainframes, none of the IBMulators dare get more than a step or two ahead of IBM in functionality, either in CPUs or disks, for fear of straying away from full compatibility; the 9370 was the first machine launched into what had effectively become an open market, and almost the only people who bought it were 3090 users with programs to distribute.
The lateness of the new IBM RTs may already have compromised their commercial success because 18-month windows of opportunity have shrunk to six months or less in the RISC world as the DECs, Hewletts, Suns and MIPS of the workstation world scramble to stay ahead of each other in price-performance: the consolidation of much of the rest of the industry around Unix System V.4, with even Foundation member Hewlett indicating that it will be adopting it leaves an AIX that is not OSF/1 – and can’t become OSF/1 for at least a year – looking more and more like a lame duck stop-gap that is not attractive either to applications developers or users.
