DASD INDEX/VTOC Sizes

[jdlowndes]

What sizes to do you use to create/initialize your z/OS DASD Volumes (3390) with for MOD3, MOD9, MOD27, & MOD54?

//PACKINIT EXEC PGM=ICKDSF,PARM='NOREPLYU'
//SYSPRINT DD SYSOUT=*
//SYSIN DD *
INIT UNIT(xxxx) VERIFY(xxxxxx) VTOC(?,?,???) VOLID(xxxxxx) NOCHECK -
INDEX(??,?,???) PURGE NOVERIFYOFFLINE

I know this is a very elementary question, but I was recently told that mine were possibly incorrect and I wondered what everyone else was using. Thank you very much in advance for your answers!

[Robert Sample]

but I was recently told that mine were possibly incorrect

Which makes no sense at all. We have disk volumes where a single database data set is placed, so a VTOC of 1 track is plenty. We have disk volumes in our storage pool where we provide 12 cylinders for the VTOC because there can be plenty of data sets on the volume. The theoretical maximum tracks for the VTOC and index, from the ICKDSF manual Appendix C. VTOC Index are 983 / 50, 2944 / 150, 9828 / 491, and 19656 / 980 for mod 3, 9, 27, 54 respectively -- but this assumes all data sets on the volume are single track data sets. 3390 volumes support 50 DSCBs per track, or 750 per cylinder -- so figure out how many data sets you expect on the volume and adjust the VTOC size appropriately. DCOLLECT tells me our site has about 16,000 data sets on 108 volumes so we average about 148 (or so) data sets per disk volume.

[steve-myers]

148 data sets per volume sort of implies 3 tracks (148/50 = 3 after rounding up), but you want more because any time a data set extends to more than 3 extents you need an additional DSCB. In Mr. Sample's example, I'd think 1 cylinder (15 tracks) would be plenty.

In my experience, it seems to me most VTOCs are grossly over allocated.