Speaking to Meta Group subscribers in the UK recently, Chris Christiansen, director of the Group’s mid-range systems strategies service discussed DEC. He began by noting that SAA-compliance, although specifically an IBM concept, is also important for DEC users because all computer users need to be vendor-independent. Christiansen believes that DEC users should view it as a guideline and write their applications to standard application programming interfaces, user interfaces and to third-party database management systems. Speaking generally about the mid-range market, Christiansen said that the mid-range computer is rapidly disappearing from the end user who will see a series of services, but not individual vendors, operating systems or machines. He believes that OS/2 will become the leading environment on the desktop,overtaking MS-DOS in 1995. However, he does not think it will be the primary server – that role will fall to Unix because of its multi-user functionality, even though Unix system management still has a few gaps leading to its choice as an operating system for the production environment as being a career-limiting one for data processing managers at the moment – Christiansen believes this will change as Unix becomes more secure.

Serious contender

He says that the VAX/VMS architecture is good on paper for fault-tolerance and transaction processing, but that DEC is currently lagging a year to 18 months behind Tandem and Stratus. He added that while DEC will never compete with these two in price-performance terms, it will be a serious contender in this market by the end of 1992 when a dedicated VAX/Rdb processor in a cluster is likely to appear. Although Christiansen is a keen advocate of third-party databases, he recognises that DEC is working hard to revamp Rdb and is going after Oracle and that, therefore, VAX/VMS users may well want to stick with or move to Rdb, so he recommends that they take their application development out of the proprietary arena and use a third-party application development environment, such as Oracle or Ingres. By early 1991, Christiansen says that DEC’s CASE tools will be available on other vendors’systems such as Sun, IBM’s AIX and so on – he also says that he is hearing good reports about the CDD/Repository that is on beta test at the moment – it works and offers good, strong base level functionality, but is not as complex or as complete as IBM’s Repository. Within two years, Christiansen thinks that DEC will make it increasingly difficult to for independent database management systems to penetrate its VAX/VMS environments by embedding DBMS functionality in the operating system, microcode or preprogrammed microprocessor. He also believes that portable Rdb will roll out in 1991 offering client functionality under Ultrix, OS/2 and MS-DOS, with server functions able to run from generic Unix, however, he admitted that DEC is experiencing a lot of difficulties in putting such a portable product together. He did not rule out the possibility that DEC was developing a new database (CI No 1,564), although he personally is unaware of such a project. As for Ultrix, he says that for users the move to OSF/1 will be costly and disruptive but will be completed by mid-1991.

By Katy Ring

Unfortunately matters are not looking so bright for DEC in the data centre. As far as Christiansen is concerned most data centres do not want two system management staffs and so despite its Enterprise Management Architecture, DEC will have to fit in with IBM, notwithstanding its deals with Computer Associates and Systems Center to develop cross-system management software. Christiansen says that it’s a political issue more than a technological one as IBM systems personnel are not going to give up systems management. Because DEC will be stymied by IBM at the data centre level, Christiansen believes that by 1993 DEC will have become a networking and software company, with the VAX being seen as a way of moving applications out from IBM mainframes to a lower cost system. The commonality between DEC and IBM system

s will be there, especially following DEC’s alliance with Systems Center for the development of Net/Master as a gateway product to NetView. Christiansen was adamant that the VAX will never manage IBM systems. There will, however, be a tighter coupling of DEC to IBM via OSI products, for DEC unlike IBM, which is building two separate stacks: SNA and OSI is tightly coupling OSI to its DECnet strategy. Although X400 is seen as the key to moving to OSI, the Meta Group is still recommending Soft Switch gateways as X400 has no proven interoperability as yet and X500 is not agreed. Fortunately DEC has a good set of SNA gateway companies – DECnet will never achieve backbone status although it is growing fast at the departmental level. This is because while DECnet is evolving in many customer sites, SNA will continue to dominate due to the its presence in the data centre. Furthermore, SNA is relatively stronger than DECnet in session management for large terminal networks – an area where DEC may require a front-end processor. Indeed, Christiansen pointed out that at the local area network level DEC is the largest reseller of Microsoft’s LAN Manager product, because it has got it to work – for 20 users and above DEC’s LAN Manager offering is cheaper and more stable than Novell’s NetWare says Christiansen. Since IBM is now attacking DEC’s peer-to-peer advantage via LU6.2, APPC and OS/2 Extended with LAN Server, users will distribute data in the IBM environ-ment. However, DEC’s Enterprise Management Architecture is likely to achieve parity with NetView by 1995, giving it plenty of room to expand at the low-end of system management. DECnet, however, is destined always to be the legbone to IBM’s SNA backbone. In this role DEC’s severest challenge will come from Tandem.

Poor stepchild

In the mid range, Christiansen believes that VMS and the OS/400 are the only two proprietary systems that will survive past the early 1990s. VMS will be enhanced so that it will have up to 2Gb of addressability, increased fault tolerance and transaction processing support, while more functions will be added so that the operating system will get richer and the database will get difficult to support and more complex. Third party vendors will have a great deal of difficulty implementing their database managment systems. As for the future of VAXes – Christiansen believes that they will be forced out of the low end by Unix RISC boxes, while although VAX transaction processing rates remain competitive the AS/400 as well as Unix boxes also offer price-performance advantages. However, VAXes will continue to shine in the technical computing market and in the graphics market, since in some cases VAX vector processing already outperforms System 370 mainframes. Christiansen said that Ultrix will remain DEC’s poor stepchild, with the DEC sales force continuing to focus on VAX, although at the low end DEC will have no option but to sell Ultrix, as Unix will dominate this area. DEC will attempt to keep the production environment the exclusive preserve of the VAX. Meanwhile, VAX prices will rise as will the cost to the user of software licen-sing to fund DEC’s penetration of the Unix market.