Probably many others will chime in on this. I have lost RAID 5 arrays with 
two disk failures within an hour of each other. RAID is nice, but one must 
allow for failures.

Long ago I was involved with reading archived tapes and transferring the 
data to CDs. The programs involved were home-written and the project ended 
up going nowhere. However, we discovered that tapes  kept too long started 
having errors. (At that point, for the CD copy, we just logged the error 
and accepted the corrupt data; what else could we do?) How long is "too 
long"?? It was variable, but measured in a few years. The advice then was 
to minimally read the tapes every year or so to "retension" them. Don't 
know if this would apply to more modern tape media.  (We also discovered 
that locally "burned" CDs are not expected last forever.)

IMHO, the key point for tape backups are (1) off-site storage, (2) 
multiple PiT recovery, (3) logical error recovery. All this can be done 
with disk-only environments involving remote copy and lots of disk space, 
but all that becomes expensive for smaller shops.

Bill Ogden


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

Reply via email to