Forget about 9370’s role as the VAXkiller, it’s better than the AS/400 in so many ways
This week, IBM’s VAXkiller, the 9370, came up for air during a day long press briefing at the company’s Bblingen laboratories in West Germany. The 9370 was launched three years ago and offered users mainframe support for Unix, Ethernet and other local area network trappings that the company had failed to deliver while promising Token Rings and SDLC loops (CI No 537). Naturally it is no more than coincidence that IBM chose this juncture (as DEC’s VAX 9000 is about to be unveiled) to declare that the 9370 is not dead. Distancing themselves as much as possible from comparisons with DEC, IBMers chose the AS/400 as the foil against which to make a marketing pitch for the 9370. DOS/VSE to have only limited SAA support The first issue that IBM addressed was the ticklish one of the life expectancy of the 9370 operating system DOS/VSE – a system beloved by 9370 users but whose existence beleaguers IBM attempts to have the semblance of a consistent Systems Application Architecture strategy. The current IBM line is that since DOS/VSE has a life support of 20,000 licences it will continue to minister to the system. According to Willi Neidow, who is manager of the Programming Centre at Bblingen, VSE is continually being renewed with version 4.1 of VSE released in June, while version 4.1.1 is up and running in beta sites now, and is due out in December. He took the traditional view that 9370 DOS/VSE users would eventually move up to a 3090 MVS/ESA mainframe. IBM claims to be working on making the migration route easier by closing the gap between the two operating systems. This was a strategy announced in the US in June (CI No 1,209) and entails improving the usability of MVS and making it easier to transfer applications from one system to another via an ESA version of DOS/VSE. Neidow stressed that IBM didn’t want simply to duplicate MVS, however, arguing that VSE must retain its simplicity and usability. In other words the upgrade path to MVS will continue to be clearly signposted although IBM denied that VSE users would be pressured into taking it. Neidow also said that VSE would never be fully compliant with SAA, but that it linked into aspects of SAA (doesn’t everything?) via the Common User Access interface. As for the future direction of VSE, Neidow said that IBM was committed to supporting the existing customer base and that following the release of version 4 of DOS/VSE, a new market was opening up for the 9370 in the distributed environment where in certain circumstances it may be a more suitable machine than the AS/400… Better than AS/400 in transaction work In certain circumstances IBM appears to be making sure, via its software development programme, that the 9370 is a more appropriate machine than the AS/400. For example IBM would not clarify whether it was working on putting CICS on AS/400, but was keen to spread the word that CICS-DOS/VSE is being worked on at the moment (although CICS/VM was not a roaring success, IBM admitted. That just about sews up the IBM departmental mid-range processor market for financial and retail transaction processing in favour of the 9370 vis-a-vis the AS/400. The other main function of the 9370, apparently, is commercial publishing. Here again IBM is being somewhat backward in coming forward with software for the AS/400 that might poach 9370 implementations for which the IBM preferred environment is (yes you guessed it) MVS. Regarding the issue of the need ultimately to merge OS/400 and MVS architectures, IBM’s unnervingly evangelical answer was: we’ve already done that through SAA. Everything is born again under Systems Application Architecture… Controller for networks of AS/400s The most intriguing argument for the longevity of the 9370 came from Penny Harrison, whose theme was that the 9370 enables IBM’s networking strategy. The idea behind this strategy is that the 9370 is the perfect network management machine for controlling clusters of AS/400s. It is to be seen as a gateway solution lin
king clusters of minicomputers to high end IBM mainframes. In this capacity the 9370, running MVS/VTAM, can just about be viewed as a feeder for the 3090 market from the AS/400 user base, or as a bridge to the 3090 market for DEC and Sun users. Either way the 9370 will take the humble role of managing distributed intelligence, running automatically or from a central site, and linked by NetView software. Ms Harrison claimed that NetView which is currently viewed as a system assisted tool will (in more than three years’ time) move to system directed analysis providing real-time mapping. This, of course, ties in with IBM’s AD/Cycle networking and knowledge systems statements of direction. Programming gap favours 9370 over AS/400 At the end of the day the director of system management at Bblingen, Dr Lennemann, stole the thunder from over-cautious, lesser IBMers by giving an extraordinarily frank review of the 9370. Lennemann’s view supported the belief that the 9370’s future lay in distributed networking, but he was also convinced that the programming skills gap would favour the 9370 rather than the AS/400. This is because you need fewer 9370s on a network than AS/400s to give the same processing power, and consequently you need fewer programmers to attend the fewer machines as well as lower level, 370, programming skills. According to Lennemann this is a valid consideration since on average 43% of a company’s data processing budget is spent on personnel. Within large enterprises 55% of 9370 customers say they chose the 9370 because they already had 370 skills, 48% bought the 9370 because they wanted standard systems to create programs and replicate them without devolving authority out to the minicomputer level, while 45% cited the 9370’s connectivity as a reason for buying it. (Some customers gave more than one reason). Smaller affiliated companies chose the 9370 for two main reasons: firstly, because of its association with large enterprise, and secondly because 370 architecture can handle both scientific and commercial requirements. This latter reason was highlighted by Lennemann who claimed that the 9370 was a better machine for mixed applications than the AS/400 or rival, Unix, offerings. As for 9370 sales, Lennemann said that 35% replace an IBM system such as a 4341, 22% replace kit from competitors, while 43% are first-time sales. He added that 42% of 9370s were sold into distributed environments, 20% to departmental environments, 33% as standalones, and 5% were classified as other. As for the future of the 9370, Lennemann revealed that all ESA architecture would run 9370 programs, that SNA and SAA support would be retained, and that the performance of high-end machines would be doubled every 18 months to two years via CMOS technology – these developments, including the pre-announced 31-bit addressing, will be upgradable. He finished by admitting to two mistakes with the 9370: the model 20 had no migration path to bigger machines, and, secondly, that the disk requirement for MVS on the 9370 was too big. This was being remedied by work on Compact Disk-Read-Only Memories, each of which would be capable of carrying more than 1Gb of data. Finally, Lennemann suggested that the successor box for the current DOS/VSE user may be a 4381 follow-on rather than a 9370 follow-on. – Katy Ring
