About a week ago we staged a DR test and found that our Novell clients restored
at an initial rate of about 5 megabytes per minute but later taper off to as
little as <1 megabyte per minute, particularly with large filesystems (around 8
gigabytes).  On some occasions I have seem the ADSM client "run out of memory"
and abend.  On my HP clients the restore throughput is essentially a flat line.
 These observations lead me to believe that the Novell ADSM client is not
managing his memory well.  In my opinion he should never run out of memory if
he has enough to start up and read in the list of files to restore.  If data is
bing logged, it should be written out to disk and purged from working memory on
the fly.  I don't see a reason that anything should build up in working memory.

My questions are:

Is anyone else seeing this behavior?

Am I being naive in my expectations of memory management (I've never written an
NLM)?

Regards,
Chuck