jmeter-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Peter Lin <wool...@gmail.com>
Subject Re: Benchmarking MS SQL Server: suggestions?
Date Sat, 12 Nov 2005 02:03:57 GMT
jmeter just uses the jdbc driver connection pool. beyond that, jmeter
doesn't know or care about how the pool works. this is assuming the jdbc
driver has built in connection pooling.

my knowledge of execution plan is limited, but tuning low level settings can
have bad side effects. if the server is tuned for specific type of query,
but different queries are sent, it could inadvertantly cause worse
performance. Sql Server does a pretty good job of optimizing queries and
execution plans, so it's generally not necessary to tune low level settings.

out of curiousity, are the queries you're running have lots of joins or
subqueries? if so, that could be a cause. one thing you might want to do is
run some basic queries to establish a base line and then work your way up.

peter


On 11/11/05, Richard Gaywood <richardgaywood@gmail.com> wrote:
>
> On 11/11/05, Peter Lin <woolfel@gmail.com> wrote:
> >
> > I'm not sure I understand your question. Stored procedures are compiled
> > when
> > it is first created with the corresponding DDL command "create procedure
> > blah...."
> >
>
> Yeah, I don't really understand it myself to be honest: the colleague who
> said that is several dozen rungs up the DBA-Tuner-Guy than I am. I suspect
> he meant that it was changing the excution plan around (the server in
> question is very highly tuned, with dozens of quite complex execution
> plans
> it picks between very carefully) so maybe that's what he meant. I'll
> follow
> it up on Monday.
>
> My original queries regarding how the database pool works in JMeter and
> the
> relative goodness of jTDS are still puzzling me though.
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message