One of the most baffling parts of the IBM fallout on September 11 was the Information Warehouse – what is it and what are its implications? Few IBMers seem to know or if they do they’re not telling, other than to say that it is a set of products and facilities to manage and deliver timely, accurate and understandable business information for effective decision-making. Since this definition could equally well apply to a post office it is not very helpful. More helpful is a release from IBM’s friendly rival Teradata, which explains the idea as follows: in simple terms, data from operational systems and other sources is fed into a central database – the Information Warehouse. This database can then service the information demands of the organisation without disrupting operational systems. The concept is simple and effective, but the word ‘warehouse’ is not really an adequate description of the interfaces, processes, data structures, controls, distribution and presentation techniques that are required to turn vision into reality. It is really an Information Factory…

Three part framework

That is the vision behind the Warehouse: how does IBM propose to architect it? Well, the framework comprises three parts: applications and decision support systems, the data delivery element, and the enterprise data element. The decision support part of the Warehouse embraces an area where IBM already has products such as its Query Management Facility, Data Extract, Data Interpretation System, Personal Application System and so on. However, it is also an area where a substantial part of IBM’s user base has third-party software. For example, it is no surprise that the SAS Institute was one of the first companies to offer support of the Warehouse concept, since the SAS System is used as a decision support system by a large chunk of the IBM user base. Of greatest significance and least surprise to the database industry is the fact that IBM has designated its Distributed Relational Database Architecture – DRDA – the data delivery mechanism for the Warehouse. Of more surprise was the news that Oracle and Informix have decided to support DRDA. Both these vendors have been extremely active in the SQLAccess Group, which is busy developing a rival standard to DRDA. So why are they supporting DRDA? The feeling of both vendors is that with IBM’s point blank refusal to join SQLAccess, DRDA is bound to become a de facto standard and vendors have to respond to user requirements for interoperability. Although IBM had publicly committed to supporting the SQLAccess RDA standard, it then started muttering darkly about having to implement both RDA and DRDA. So it seems that the rest of the database industry is having to compromise and will tweak RDA to create a gateway to DRDA. As part of the announcement IBM said that phase 1 of DRDA – the Remote Unit of Work level – will be available for DB2 and OS/400 in March next year, for SQL/DS in May 1992 and offered a statement of direction that OS/2 will be a Remote Unit of Work client. Third-party database vendors fully supporting DRDA can also be connected. The Remote Unit of Work level means that, for example, a TSO user can point to a database on VM and run SQL statements against it, but can’t access databases on VM and MVS at the same time, as two tables cannot be joined across two different databases. Furthermore, the user has to point the statement as there is no location transparency. So what do third parties think of IBM hijacking the relational database standards process via the Information Warehouse? Oracle’s John Spiers thinks the coup can be viewed two ways: firstly, IBM is simply seeking to control what it couldn’t stop third parties doing anyway – that is accessing IBM databases via SQL; secondly DRDA makes it far simpler for third parties to connect their databases to IBM’s.

By Katy Ring

So is the Information Warehouse really a kind of ‘Freedom of Information Act’ for IBM customers as TechGnosis claims? If pushed, IBM UK’s database product manager John Cole will admit that Oracle or any o

ther third-party database can, at the moment, be the central database within the Information Warehouse, but evidently thinks that over time maintenance and building tools will be developed that will make DB2 the natural choice for the operational database at the centre of the Warehouse. However, relational database squabbles pale against the real import of the announcement, which was to promote the use of SQL as the common language to access non-relational data – that is about 90% of existing data. This takes us neatly to the Enterprise Data part of the announcement where Information Builders Inc, an IBM International Alliance Member for the Information Warehouse, has a starring role with its Enterprise Data Access/SQL family of four client-server products. These are EDA/SQL Server, a host component that processes SQL requests against relational and non-relational data; API/SQL, which is a call level application programming interface; a modular system of communication interfaces called EDI/Link and EDA/Extender Products, which are direct interfaces from SQL-based tools. All in all this adds up to a very sophisticated information retrieval system offering SQL access to 30 different operating system formats and 45 different types of database interfaces. Information Builders is publishing the EDA/API specification as an open standard and is selling a software development kit for large software houses to use to make their products EDA-compliant by installing EDA extenders. In this way existing applications can be modified to take part in the EDA technology and small software houses can ask Information Builders to put the API in their application for a small fee. So in order to partake of the Information Warehouse concept, a vendor simply has to ensure that its products are, as appropriate, EDA certified and/or DRDA-compliant.

Astounding conclusion

The one astounding conclusion to be drawn from all this is that a user can implement the Information Warehouse framework without buying any IBM kit at all – the mainframe server and the distributed Unix workstations do not have to be IBM’s and neither do the databases, tools and applications. Of course, IBM expects the mainframe server to be a 3090 with MVS and DB2 and the distributed hardware to be AS/400s and PS/2s – but there is nothing in the Information Warehouse to force the user to make those decisions. IBM will tell users they need DB2 to implement the Warehouse, frankly speaking they do not and any intelligent user will evaluate carefully what IBM, Teradata, Oracle and Software AG et al have to say before deciding which operational database will form the centre of their strategy. It is also worth pointing out that Information Builders’ EDA technology is available for, the admittedly still small, world beyond IBM. Users opting for a DEC, Bull or AN Other strategy can make use of EDA. In all, there appear to be three main conclusions that can be drawn from the Information Warehouse: firstly it is IBM’s bid to ensure that the Enterprise server is a mainframe using a relational database; secondly, the user has now been given a charter to use any third-party software it wants; thirdly, the Warehouse means that Unix and non-relational databases are now acceptable as part of SAA even if they don’t yet officially make the AD/Cycle grade.