Legacy Databases Take a Dive

A close look at the TDWI-Forrester Quarterly Technology Survey of 290 data warehouse (DW) users shows that there are multiple winners in the dynamic and growing market for data warehousing based on the dominant design of the relational database. In the 18 months between our last survey in May 2003 (not shown) and the August 2004 survey (see Figure 1):

  • Oracle has increased its installed base from 42% to 44%.
  • Microsoft SQL Server has increased its installed base from 19% to 21%.
  • IBM's DB2 has increased its installed base from 14% to 18%.
  • Sybase has increased its installed base from 1% to 2%.
  • MySQL and PostgreSQL - open source database management systems (DBMSs) - are now on the radar, registering a combined total of just less than 2%.
  • Teradata continues to hold its own at 8%.

Figure 1: Database Platforms for Production Data Warehousing

The losers? Legacy databases have declined from 12% to 3%, and Informix decreased from 3% to 2%. IBM's DB2 picked up an extra percentage point from Informix, growing by 3%, whereas, as indicated, Oracle and Microsoft SQL Server grew by 2%. In spite of holding an installed base best described as "dominant," Oracle continues to face intense competition in high-end data warehousing - not only from Teradata, but also from upstarts such as Netezza and Sybase's IQ Server. Likewise, in the Windows market, Oracle faces not only Microsoft, which has been hurt noticeably by the delay in shipping the latest version of SQL Server, but by IBM's DB2. Oracle will benefit in its battle for installed base against Microsoft by the Unisys/Oracle data warehousing benchmark, in which Unisys used Oracle10g, not Microsoft SQL Server, in its 64-bit, 16-way ES7000 Enterprise Server configuration. Notwithstanding the many assurances from Unisys that Microsoft is a valued partner, it's not hard to read between the lines - SQL Server does not scale as well.

Potential Vulnerabilities

As the overall leader in the unsegmented data warehousing market, the game is Oracle's to win or lose. Oracle has strength, but potential vulnerabilities include:

The paradox of high availability. In its earlier incarnation, Oracle's RAC showed what might be described as the paradox of high availability (HA). The more moving parts that are added to maintain system availability, the greater the likelihood that, in the worst case, one of them will fail, resulting in the very scenario against which the HA was supposed to be the antidote. In the best case, end-user clients reported that the administrative overhead was significant enough to give pause in making the adoption decision. Oracle has diligently improved RAC, and early reports from 10g users are that clients who would not engage the earlier version are making a commitment.

Oracle's business practices. In particular, for those clients who operate PeopleSoft data warehousing options such as Enterprise Performance Manager (EPM), Enterprise Data Warehouse (EDW) and Enterprise One, the future is uncertain. The list of vendors that have been acquired whole by Oracle includes Carleton (for data quality), OneMeaning (for meta data), and Darwin (for data mining). The slogan for the legacy clients was "You don't need any other vendor besides Oracle." In reply, Oracle points to the examples of the DEC relational database, RDB, which Oracle purchased from DEC years ago, as well as Oracle Express, as examples of long-term, high-margin support of legacy products. The issue is how comfortable users are with a functionally stabilized feature set.

Oracle's Daily Business Intelligence (DBI). In DBI, aggregates within the transactional application are highly normalized and express important key performance indicators, provided, however, that the answer lies completely within Oracle11i applications. Oracle claims that this addresses performance problems within the ERP system by leveraging RAC to put query and summarization on its own node. Meanwhile, materialized views can be the target of high-performance, incremental loads, decreasing strain on 11i. The result of relying on aggregates derived directly from the underlying transaction system is that the client gets a data mart-class solution without the mart.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access