Hey Alan, and anyone interested.

You say you peak out at about 190-200 TPS at peak. Is that a per disk
number, filesystem or database period? I'm trying to get to the
bottom of this serial versus parallel database access thing, and I'm
having a hard time doing it. As I mentioned, if TSM only does serial
access to the database, then with the biggest, baddest array in the
world with an average latency of 3ms you're only going to see ~300 IO/
s per second. So, basically, we may as well put the databases on a
couple of IDE drives with a 3Ware RAID controller from Fry's and IBM
can keep their monster arrays they are selling to everyone 'cause
"the database needs spindles".

Thoughts?


Jason



On Oct 18, 2006, at 7:43 AM, Allen S. Rout wrote:

On Tue, 17 Oct 2006 12:23:36 -0400, Sergio Fuentes
<sfuentes AT UMD DOT EDU> said:

- nmon traces and the nmon_analyzer are really good at
illustrating your overall
performance.  topas, or top is also useful for active monitoring.
vmstat as
well.  During our slowdown our DB was seeing an average of 30-40
TPS per disk.
Post-slow down we now see 90-120 TPS per disk.

So what stanzas do you use for data collection and analysis here?

I think it would be immensely useful if we could compare similar
results.

I see 15-40 TPS in normal backgroun work, and peak out at 190-200, and
I've got 10K 18G SSA, usually with 2 or 3 LVs on each PV.  But it's
hard to compare, because I've got lots of servers sharing the same DB
volumes.

- Allen S. Rout

--
Jason Lee
DreamWorks Animation
(818) 695-3782