What script or tool do you use to benchmark a server for OLTP? I'm
trying to decide on the best RAID and fragmentation strategy for a new
system and I need something that can simulate a large load coming from
many users (300). I've had a look around and found a few free or
open sourced tools for other database server vendors but so far nothing
for Informix. Or is there anything homegrown that you can share?

Thanks,
Jim

Check Server Studio JE. It has got a Benchmark Runner. I haven't had
chance to use it.
http://www.serverstudio.com/

hariog@yahoo.com wrote:
> Check Server Studio JE. It has got a Benchmark Runner. I haven't had
> chance to use it.
> http://www.serverstudio.com/

I should have mentioned that I had a look at SSje, the version that is
bundled with IDS only allows up to 5 sessions in total, and you also
need to write all your own tasks/scenarios.

Doesn't anyone benchmark their servers before deploying them or do you
simply use a best guess?

Jim Kenedy said:
> Doesn't anyone benchmark their servers before deploying them or do you
> simply use a best guess?

Why bother? No-one is going to listen to you anyway.

--
Bye now,
Obnoxio

"... no bill is required as no value was provided."
-- Christine Normile

--
This message has been scanned for viruses and
dangerous content by OpenProtect(http://www.openprotect.com), and is
believed to be clean.

Obnoxio The Clown wrote:
> Jim Kenedy said:
> > Doesn't anyone benchmark their servers before deploying them or do you
> > simply use a best guess?
>
> Why bother? No-one is going to listen to you anyway.
>
True, but I'm in the position of not having anyone listen to me since I
have carte blank over these servers. I simply would like to get every
ounce of performance out of these servers. I've looked at some of the
TPC-C results and I've been thinking that it would be good to have a
standard benchmark test that all users can use to run against their
systems. And by sharing our results we would have a better idea if our
systems are performing optimally. There are a myriad of ways to
configure hardware and software nowadays and it would be nice if such
tools were more available for the average Joe.

> > > Doesn't anyone benchmark their servers before deploying them or do you
> > > simply use a best guess?
> > Why bother? No-one is going to listen to you anyway
> True, but I'm in the position of not having anyone listen to me since I
> have carte blank over these servers. I simply would like to get every
> ounce of performance out of these servers. I've looked at some of the
> TPC-C results and I've been thinking that it would be good to have a
> standard benchmark test that all users can use to run against their
> systems.

My opinion is that, in most real world cases, these standard benchmarks
are basically useless. What matters is how fast your application runs;
you should build site specific benchmarks based upon your site specific
applications. The best way - use the applications to do the
benchmarking.

If you can script your applications some how, do that.

> And by sharing our results we would have a better idea if our
> systems are performing optimally. There are a myriad of ways to
> configure hardware and software nowadays and it would be nice if such
> tools were more available for the average Joe.

Database engines aren't like "engines"; you can't measure performance
in absolute terms like horsepower and torque-pounds. Tuning up for one
type of query (including one used by a benchmark) might be tuning down
for another type of query. It matters allot what kind of queries you
actually use.

Jim Kenedy said:
> Obnoxio The Clown wrote:
>> Jim Kenedy said:
>> > Doesn't anyone benchmark their servers before deploying them or do you
>> > simply use a best guess?
>>
>> Why bother? No-one is going to listen to you anyway.
>>
> True, but I'm in the position of not having anyone listen to me since I
> have carte blank over these servers. I simply would like to get every
> ounce of performance out of these servers. I've looked at some of the
> TPC-C results and I've been thinking that it would be good to have a
> standard benchmark test that all users can use to run against their
> systems. And by sharing our results we would have a better idea if our
> systems are performing optimally. There are a myriad of ways to
> configure hardware and software nowadays and it would be nice if such
> tools were more available for the average Joe.

I won't disagree with you. The problem is, that unless you get your
putative benchmarking tool certified by some group or t'other, it won't be
worth a hatfull of dandruff.

--
Bye now,
Obnoxio

"... no bill is required as no value was provided."
-- Christine Normile

--
This message has been scanned for viruses and
dangerous content by OpenProtect(http://www.openprotect.com), and is
believed to be clean.