On 5/11/2016 7:31 AM, Vernooij, CP (ITOPT1) - KLM wrote:
I heard rumors that VIO is that CPU expensive and that I/O is that fast nowadays, that it is VIO is not worth the extra CPU anymore.
We use VIO for (nearly) ALL temporary data sets. (VIO MAXSIZE is set to the highest supported value of 2G.) I haven't tried to compare VIO against DASD from a performance standpoint -- probably because we never noticed any issues with VIO.
For us it's a matter of convenience to have numerous, fairly large "DASD" work data sets that never, Ever, EVER experience VTOC data set naming conflicts or need scratching after a system failure. Another benefit to us is the ability to have a virtual 3380 device for testing code intended to be "DASD-geometry-agnostic" but might not be.
We recently uncovered an error with IBM's stand-alone ADRDSSU because we used a temporary 3380 device in the SABLD job. The error is not specific to 3380, it just happens to be reproducible right now on 3380 given the current module sizes. If the error is not fixed, the same problem could occur on 3390 if some PTF changes the sizes of the load modules just right...
-- Edward E Jaffe Phoenix Software International, Inc 831 Parkview Drive North El Segundo, CA 90245 http://www.phoenixsoftware.com/ ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
