Re: Mainframe near history (IBM 3380 and 3880 docs)

Long, long ago, I did part of a Master's thesis using a data cell.  (It
involved writing a 2260 full-screen editor and library management routines
to store source programs on the data cell.) We found the trick was to
avoid
PDS data sets and OPEN functions.  Both these concentrated access on the
strip(s) containing the VTOC or the PDS directory.  By using BDAM and
storing our "directory" on normal disks, we would access only the data
cell
strips containing the desired "mini-files".  We had very few crashes.  In
our simple analysis it appeared that much of the data cell crash problem
was
related to heavy use of just a few strips.

Bill Ogden