PAC Brion Arnaud wrote:
>
> Your instructor should read his books again ...
The question wasn't about migration _processes_, but migration _threads_
internal to TSM. As you and Andrew both describe, there are multiple
migration _processes_ kicked off in both cases.
I hadn't heard anything about the "migrate stgpool" command running the
migration processes all under a single thread (I don't know if that is
even possible). So I looked at the (unsupported/undocumented) command
"show threads" to see what I could find.
I currently have 6 migration processes running, 4 initiated by reaching
the migration threshold and 2 initiated by "migrate stgpool". Looking
at the "show threads" output, I see 6 DfMigrationThread thread running.
I checked earlier when no migrations were running and saw no threads of
that type.
So at a glance, it would seem the instructor is in fact incorrect. But
if any TSM internals expert would like to offer a definitive answer,
that'd be great. It would be unfortunate if "migrate stgpool" really
was a lower performance method.
> -----Original Message-----
> From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf
> Of
> Meadows Andrew
> Sent: Tuesday, 27 November, 2007 21:01
> To: ADSM-L AT VM.MARIST DOT EDU
> Subject: Mig stg question.
>
> Hello list.
>
> Environment .
> TSM 5.3.4.1 running on AIX.
>
> I have a question about the mig stg command versus the hi/lo parameter
> on a Disk Storage pool for migration purposes.
>
> I was recently in a TSM class and the instructor informed me that the
> mig stg command runs single threaded as compared to the upd stg command
> which will run multiple threads.
>
> Here is the theoretical situation I gave him.
>
> You have a disk pool. That is 10 GB. This disk pool has the Migration
> Processes parameter set to 2.
> Theoretically this disk is 50 % full with data from 2 nodes. And tyou
> have the tape drives available for migration to the Next Storage Pool to
> tape.
>
> Scenario 1.
>
> You issue the mig stg diskpool lo=0.
> This will kick off 2 processes.... 1 per node....
> According to the instructor this is running on a single thread and
> moving the data at a slow rate.
>
> Scenario 2
>
> You issue the upd stg diskpool hi=0 lo=0 This will kick off 2 processes
> .... 1 per node.....
> According to the instructor this is running one thread per process and
> moving the data in paralell.
>
> Has anyone seen this, or heard this? We are currently using the mig stg
> command and I wonder if we could be doing this more efficiently.
>
> Thanks,
>
>
> Andrew Meadows
> Storage Services Operations
> Enterprise Storage Services
> HCA Information Technology & Services
>
--
Hello World. David Bronder - Systems Admin
Segmentation Fault ITS-SPA, Univ. of Iowa
Core dumped, disk trashed, quota filled, soda warm. david-bronder AT uiowa
DOT edu
|