Some people are over-invested in old technology
- IT TOPICS:Business Intelligence, Management, Software
Alf Pedersen has made a blog post questioning my column introducing the DBMS2 concept. My editor thinks that this “calls for a response.” So I’ll try to summarize a few points Alf is missing.
- Almost no enterprise with multiple application software vendors can ever have a wholly consistent enterprise-wide data schema. Almost every large enterprise has many application software vendors.
- Mergers happen, and when they do they play merry hell with enterprise schemas.
- Data warehousing represents a rapidly growing percentage of total data management costs, at the expense of OLTP, and almost all the coherent arguments for a single-database strategy are focused on OLTP.
- Text management is increasingly important.
- Market-leading DBMS carry huge baggage because of the relationships among various categories of hardware costs at the time they were designed. They still do a great job for many applications (including some important new ones) – but for an increasing number of other apps, they are useless or at least obsolete.
- Modern pointer-based data management faces all the same challenges that IMS and IDBMS failed at 20-30 years ago. But 20 years is a long time, and those problems CAN be solved. The best example today is Cache’, but more XML-centric solutions are coming fast (e.g., IBM’s “Viper”).
- SOAs have won. SOAs are XML-based.
________________________________________________________
FROM THE EDITOR:
This blog post has been edited.
Additionally, this comment thread has been closed, in line with Computerworld's Terms of Service. Computerworld wants to foster a civil and respectful debate over important IT issues, but this thread has become too personal and not useful to Computerworld's audience of IT professionals. Certain comments may be reposted at a later date, but new comments will be disabled.
Ian Lamont
Online Projects Editor
Computerworld.com
ian_lamont@computerworld.com



