Re: CODASYL-like databases
Date: Tue, 1 Apr 2008 11:47:57 -0700 (PDT)
While this passes as received wisdom, I can't find any actual evidence to support it. At least, nothing since about 1992.
When I talk to folk who make and sell disk systems and what-not, they tell me that they for every 1G of disk they're selling under IMS/ network systems, they're selling 100G for "relational" systems (largely warehouses), and 1000G for file-systems (emails, and attachments).
When I talk to folk who try to sell services to IT departments they tell me that the overwhelming preponderance of "business data" today is in Excel Spreadsheets (quotes, models and so forth) and the majority of "business transaction data" is being managed by RDBMSs.
Very, very few of the web based applications developed since 1995 have used IMS.
It's absolutely true that there are still some very large IMS systems out there but their number, and the amount of data they manage, doesn't seem to be growing. From what I can tell much of the data they hold is short-codes-and-numbers stuff, while modern businesses are increasingly dependent on other kinds of data; transcriptions of support calls, customer feeback text, and so forth.
If you define "business data" extremely narrowly - essentially to just buy/sell/ship/receive stuff - then it's remotely possible that IMS still dominates (though I doubt it). But if you include all the personnel records management, sales management, and what-not, no way. Received on Tue Apr 01 2008 - 20:47:57 CEST