I'm currently monitoring a 'backup stgpool process that has been
running for the past 24.5 hours. We've now copied
116,651,182,857 bytes (roughly 108GB) of data from the primary
pool to the copy pool. Working that out, the transfer rate
between tapes is a bit more than 1MB/s (~1.3MB/s).

During the course of the backup, 24 primary storage pool tapes
have been mounted.

We are using 3590 drives in a 3494 library. Each drive is on
its own SCSI bus. Our ADSM server is a wide node of an SP2, with
256MB of RAM. The database size is 22GB.

I'm curious to know why we might be seeing such a low transfer
rate. The 3590s are supposed to be rated at 9MB/s, so during
regular sustained operation I'd hope to see at *least* 3-4 MB/s.

I just realized that our dsmserv.opt has the values of
MOVEBATCHSIZE and MOVESIZETHRESH switched around (MoveBatchSize
is set to 500, and MoveSizeThresh is set to 256, while it should
be the other way around). Would having an invalid value for
MoveBatchSize (500) set the value to the default (40) or to the
max (256)?

Any thoughts or suggestions of where else we might look to
improve our performace are appreciated.

 -- Tom


"If I could dot the 'i' in a Michigan     Thomas A. La Porte
 game and the good lord came to take me   DBA, Feature Animation
 the next day ... at least I could        DreamWorks SKG
 die happy." - Beano Cook, ESPN           <tlaporte AT anim.dreamworks DOT com>