When running backups in our SAN our backups start off running @27MB/Sec.
But as time goes by this eventually will drop all the way down to about
11MB/Sec.  However, when I am running I restore it is able to maintain the
@27MB/Sec.

We have tried several different tunings of buffer sizes and multiplexes to
no avail.

Currently I am running:
NUMBER_DATA_BUFFERS = 64
SIZE_DATA_BUFFERS = 256k
MPX=6

I have set fragmentation size between 0 and 2048 w/o any change in
performance.

Based on entries in bptm log I appear to be underrunning the drives:
     write_backup: waited for full buffer 15893 times, delayed 34674 times

But when I am performing restores I don't see these large numbers:
    mpx_read_data: waited for empty buffer 2 times, delayed 30 times

This particular SAN Media Server is our 'new' development server and is an
HP L3000 - 4CPU/550mHz w/ 4GB RAM.  It isn't getting a lot of action yet so
I would think that the server would be beefy enough to push the drives.


Thanks in advance!
Eric


-- 
Eric Buckner 
Dollar General Corporation 
UNIX System Administrator 
Voice: (615) 855-4381 
Email: <mailto:ebuckner AT dollargeneral DOT com> 
Quote: "Time is not a test of the truth."
Translation: Just because you've always done it that way, doesn't make it
right.