Hi Jack -

For many, many years we had been iteratively testing the Erase-On-Scratch 
function to determine if the performance had improved over time (with new 
hardware/software releases) to determine if we could exploit EOS, but without 
getting too far into the weeds, the bottom line was that the overhead to 
implement EOS was too great--specifically the elapsed time.  IBM did great work 
to dramatically improve the performance over time, but there was always going 
to be additional CPU and EXCP overhead--no way to avoid that.  But we 
determined that the elapsed time it would take to scratch some of our larger 
data sets would negatively impact our batch window, so we never implemented it. 
 Yes, you can select certain data sets instead of using "ALL", but that 
involves too much administrative overhead (and continuous updates as the 
environment changes) and at the end of the day, some of the largest data sets 
probably have the most sensitive data in them.

That said, you might want to check out IBM APAR OA61492 (and associated PE 
fixes).  The DASD UNMAP function (or the non-IBM DASD equivalent) seems to 
perform "the same" Erase-On-Scratch function... but at the hardware level.  We 
are using the EMC equivalent and it is working seamlessly--no z/OS side 
overhead at all and practically nothing at the DASD hardware level.

Larre Shiller
US Social Security Administration 
“The opinions expressed in this e-mail are mine personally and do not 
necessarily reflect the opinion of the US Social Security Administration or the 
US Government.”

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to