-Although AT&T and Unix International appear to be winning the battle to establish the core standard Unix operating system, the rival Open Software Foundation has already proved its abi lity to push companies into working together on some of the surroun ding technologies with the other OSF Motif user interface – and is now beginning the whole process again for distributed computing. John Abbott reports

Late last month, the Open Software Foundation released the names of some 30 companies and organisations that had responded to its distributed computing Request For Technology, issued back in June. The Request, which called for distributed computing environment technologies that enable users to share resources (such as CPU power, data and peripherals) over a network, drew an initial 50 responses, although only 28 were judged to meet the required criteria. The Foundation said that the Request was unique in the number of companies that chose to respond with joint submissions, something which it said would ultimately benefit the entire industry by producing superior technology. Four joint submissions were witheld while the companies concerned prepared joint announcements. Initial indications, however, appeared to indicate that the battle would be more complex than the predicted two-pronged fight between the most dominant technologies, widely recognised as Sun Microsystem’s Open Network Computing and Hewlett-Packard’s Network Computing System, developed by its Apollo Computer Division.

Hedged their bets

Indeed, most of the major players appeared to have hedged their bets with various options, while others submitted only specific technologies that might be combined with other submissions. IBM, for instance, currently uses Network File System for its AIX systems and mainframes, while it has licensed Network Computing System from Apollo, and appears to be involved in a third submission to the Foundation, through the Transarc Corp, in which it has a financial stake. Despite widespread endorsements of its Open Network Computing technology following its joint announcements with Novell Inc and Netwise Inc recently, Sun chose not to be a part of a joint announcement, submitting only part of its Open Network technology, and simply pointing to the complementary submission from Novell Inc. Sun’s offering was the transport-independent portion of its remote procedure call, RPC XDR (external data representation), which it jointly developed with AT&T for inclusion in Unix System V.4. RPC XDR works in conjunction with the Transport Layer Interface in Unix V.4, which according to Sun spokesman Dennis Freeman will enable TCP/IP, NetWare, and eventually Open Systems Interconnection protocols to be implemented on distributed systems, with applications not needing to know the transport mechanism used. RPC XDR is already used in Network File System, currently licensed by around 280 vendors, and support from Novell, Banyan and 3Com opens up the world of personal computers. Sun is leaving aside its own higher level networking tools in favour of the RPC Tool compiler from Netwise Inc – called by Sun’s vice-president of distributed computing Larry Garlick the Adobe of distributed computing (he means PostScript) – which being independent of both Sun and AT&T should strengthen its case.On the face of it, however, the Apollo Network Computing System, which has for a long time been regarded as a technically elegant solution for distributed computing despite its long gestation period, has major backing amongst the Open Software Foundation’s sponsors. Over the past year DEC, Hewlett-Packard, IBM, Dell Computer and others have licensed the technology, and according to Computerworld, Microsoft Corp is now also planning to endorse the NCS remote procedure call to run over OS/2, AIX Version 3 and Hewlett-Packard’s translation of the OS/2 LAN Manager into C, known as LM/X. The paper says that IBM and Microsoft – both included in the Foundation list – have combined to submit extensions to Network Computing System that will enable this to happen. IBM also ap

pears to have submitted technology in conjunction with its long term communications partner Locus Computing, along with Hewlett-Packard and Pittsburg-based Transarc Corp. Transarc recently announced that its AFS 3 version of Carnegie-Mellon’s Andrew File System will be available in March (CI No 1,294). AFS 3.0 is a very scalable system, according to Transarc’s director of marketing Phil Lehman, who claimed that AFS could scale to thousands of users at large sites, but still improve productivity in workgroups with a small number of computers. On AFS systems, such as the 10,000 user, 17 file server and 2,000 client machine system currently based around Carnegie Mellon University in Pittsburgh, users see a single large file system, and do not need to worry about the location of files. Currently working on a translator to provide interoperability between AFS and Network File System remote calls, Transarc could presumably do the same for Network Computing System if it was chosen as a portion of the Foundation’s final product. Other submissions to the Request included a separate entry from Banyan Systems, Westboro, Massachusetts, proposing its distributed naming system, providing a unified naming space for network resources. BBN Communications Corp of Cambridge, Massachusetts, couldn’t be reached to explain its submission, but pioneered Internet a decade ago, and has expertise in TCP/IP and network management. But amongst the submissions from DEC, Hewlett, IBM, Locus, Microsoft, Netwise, Nixdorf, Retix, Siemens, Sun and Transarc were those from smaller, less known companies, including Architecture Projects Management, Bit31, Chemical Abstracts, Child Systems, Paris-based Chorus Systems, Database Engines, Dialogue Switching, DEC, Dset Corp, Norcom, Research Triangle, TRW, Uniware and VXM. University submissions were received from Cornell, two research projects (Mercury and Athena) from the Massachusetts Institute of Technology, and from York University in the UK.

Sun or Apollo

Presentations by all participants are set for the five day membership meeting in Boston, from November 6 to 10. The Request evaluation team has been established from permanent Foundation staff, sabbatical employees, consultants and analysts, and will make a final selection in the second quarter of 1990, combining elements of the submissions to make a complete offering. But unless the Foundation makes an unexpected and clean start, it looks inevitable that, as the fundamental Procedure Calls for the Sun and Apollo systems are incompatible, a choice of either one or the other has to be made, whatever else is chosen for the higher levels.