To follow up on this topic -- in many cases, it was a case of "the tail wagging the dog" -- the requirements of DB2 largely led to many of the features in MVS/XA, and DB2 was the first product to really attempt to "push the envelope" and exploit those features.
In particular, DIV, or "data-in-virtual," was added to MVS, and LDS (Linear Data Sets) were added to VSAM in MVS/XA, specifically for DB2. (You have to use LDS with DIV). And, since DB2 for MVS was developed at San Jose Research (later Almaden), this is also where the original hard disks were developed. So part of the motivation for DB2 was to "sell more disk drives" and more and bigger mainframes. As the emphasis of DB2 was on end-users being able to create their own database applications, without requiring the typical "glass house" processes of extensive requirements -> design reviews -> coding -> unit testing -> integration testing -> delivery (the so-called "Waterfall model"), part of the goals of DB2 was to drive further demand for end-user computing and the "improved productivity" that resulted, thus the need for more CPU, more disk, etc. As DB2 continued to "push the envelope" and as Jay mentioned, MVS/XA was not quite up to the heavy demands of DB2, early on, so the DB2 tail once again wagged the MVS dog, leading to the development of MVS/ESA, with Access Registers and the ability to access multiple address spaces, data spaces, hiper spaces, etc., thus enabling much larger virtual storage for database buffers, code, etc. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
