Re: [Networker] Group parallelism
2008-03-17 13:48:49
On 17 Mar 08, at 1431, Keith Roy wrote:
We are running Networker 7.3.3 on Solaris 2.8 and are seeing our Group
Parallelism misbehaving. We have 10 SDLT drives with the device
target
setting of 4. If we have a group setup with a parallelism of 8,
only two
drives should be used.
Is that necessarily true? What is there to stop the group being run
with one stream going to each of eight tape drives?
I can imagine scenarios where that's pretty much the right thing, too:
towards the end of the backup window, you can end up with most of the
drives running, but taking less than their target parallelism. I have
six LTO3 drives, and I routinely see scenarios where each of them is
taken one or two streams (with a target parallelism of four): lots of
jobs kick off in the 8pm--11pm timeframe, and by 4am a lot of them
have finished, with those that are still running randomly distributed.
If a group permitting (and able to drive, given the number of clients
and the client parallelism) eight jobs then starts, why shouldn't it
send a stream to each of the drives that has capacity?
Imagine there's four drives. At 8pm, sixteen jobs start. By 2am,
half of them have finished, with 3,3,1,1 streams on the respective
take drives. At 2.01am, a 8-way job starts. What do you think should
happen?
If you want to limit the jobs to two drives, you could limit the pool
the jobs go into to two devices. But that has other consequences, of
course.
ian
To sign off this list, send email to listserv AT listserv.temple DOT edu and type
"signoff networker" in the body of the email. Please write to networker-request
AT listserv.temple DOT edu if you have any problems with this list. You can access the
archives at http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
|
|
|