As a "Rule of Thumb" (and thumbs come in various shapes and sizes ... yours may vary):
1 PLPA dataset 1 COMMON dataset 4 (or more) LOCAL datasets LOCAL datasets should peak at no more that 30 (or so) percent utilization, else the block paging algorithm starts to have problems (the probability of finding a block with sufficient contiguous free slots drops below .5) With DYNAMIC (only) PAVs, you may be able to get away with 2 LOCAL page datasets, since you will get 2 addresses assigned to each dataset Best to have the PLPA and COMMON datasets on separate volumes, sufficiently sized. Next best (for a no-PAV environment) is a 1 cylinder PLPA and a COMMON dataset sized somewhere around 2GB (2000 cylinders is probably OK). With PAVs, we need someone else to chime in here, since you probably don't want to assign 2 PAV addresses to a 1 cylinder PLPA dataset! Note that HYPER-PAVs don't play in the paging subsystem yet. Still awaiting word on this issue from IBM (Jim Mulder, maybe?). ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
