Mailing-List: contact jmeter-user-help@jakarta.apache.org; run by ezmlm
From: "Mark McWhinney" <msm@portata.com>
To: "'JMeter Users List'" <jmeter-user@jakarta.apache.org>
Subject: RE: Memory / CPU usage
Date: Wed, 3 Sep 2003 11:17:39 -0700
Organization: Portata
Message-ID: <000f01c37247$a965c750$0201a8c0@msm>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="US-ASCII"
Content-Transfer-Encoding: quoted-printable
Importance: Normal
In-reply-to: <20030903173135.89810.qmail@web13307.mail.yahoo.com>

It would be interesting to see the difference.

I will send the jmx to you directly.

-----Original Message-----
From: peter lin [mailto:jmw00lfel@yahoo.com]=20
Sent: Wednesday, September 03, 2003 10:32 AM
To: JMeter Users List
Subject: Re: Memory / CPU usage


if mark is willing to post a description of the test plan he used, I'm
willing to run OptimizeIt on JMeter 1.9.x to see anything obvious pop =
out. I
have a license of it, so running it is fairly simple.
=20
peter
=20


mstover1@apache.org wrote:
Despite what most people think, Java is not particularly slow. It is a
memory hog. If 20=20
threads is causing your JVM to use 350MB memory, you may have to take =
steps
to reduce=20
your memory usage.=20

1. Smaller number of test elements. Every element is cloned for each =
thread.
If you have=20
500 elements and 50 threads, that's 25,000 objects. A lot of people want =
to
make perfectly=20
realistic test scripts, but I'd recommend going for something more =
abstract
and really=20
reducing the number of different requests made.

2. Watch the listeners you use. Do not use a View Results Tree. If you =
have
memory=20
problems, try using only the Aggregate Report and see if it helps. You =
can
always store the=20
results and have a look at them in a different listener later.

I run 50-100 threads all the time without much difficulty. I'm not =
stressing
my CPU at all with=20
that (P4 2Ghz), and I can stay under 200MB of RAM usage. I do sometimes =
run
into IO=20
bottlenecks, however.

Incidentally, using GCJ is quite likely to give you a slower app, not a
faster one, and even if=20
JMeter is one of those few apps that sees a 10-15% benefit from GCJ, you
will never notice=20
that difference - particularly since JMeter spends so much of it's time
waiting on blocking IO=20
calls.

-Mike

On 3 Sep 2003 at 10:06, Duncan Frostick wrote:

> That's Java for you. As great as the portability and libs are, it's =
such=20
> a painfully slow lang for performance tasks. I have a Solaris server=20
> with 2GB of RAM and it can't do 100 threads without giving me stupidly =

> high response times (of close to an hour...) on some requests.
>=20
> Its a pity because JMeter is such a fantastically good tool when it=20
> works, its just let down by performance issues. If you want to load =
test=20
> at anything like a high load with JMeter you need several high powered =

> servers doing it in unision to spread the threads.
>=20
> I was trying to use the GCJ compiler - which claims to compile java=20
> bytecode to machine code - to increase performance but I couldn't get =
it=20
> working, that could be a possible performance boost. My general rule =
of=20
> thumb is that for anything above 75 users JMeter/Java is too resource=20
> hungry to give accurate results on my hardware.
>=20
> And now I'm trying high-loads, thats a problem. However, IBM have a=20
> nifty if hard to configure tool called 'stress'. It's written in C so =
it=20
> performs very well, just lacks a lot of JMeters great functionality. =
You=20
> can find it online if you serach for IBM Web Performance Tools.
>=20
> Perhaps a port of JMeter to C++ should be considered? I know this =
would=20
> involve doing a lot of painfully boring networking and threading code=20
> but the performance gain would be huge.
>=20
> Cheers, Duncan
>=20
>=20
> Mark McWhinney wrote:
>=20
> >Hi all,
> >
> >I am a long time LoadRunner user but am new to JMeter. I am impressed
with
> >its functionality. It is a lot closer to LoadRunner than many of us =
in
the
> >LoadRunner community realize.
> >
> >I am going through the usual newbie issues but am actually reading =
the
> >manual and figuring it out.
> >
> >I ran a test with one thread group with 30 threads that had six HTTP
> >requests with a 5 second wait time between each request. I maxed out =
the
> >CPU at about 20 to 25 users on a 800 MHz Windows 2000 machine. Memory
> >utilization was about 350 Mbytes. Is that typical? I am really going =
to
> >need 10 machines to do a 250 user load test or am I missing =
something?
> >
> >
> >---------------------------------------------------------------------
> >To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
> >For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
> >
> >=20
> >
>=20
>=20
>=20
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
>=20


--
Michael Stover
mstover1@apache.org
Yahoo IM: mstover_ya
ICQ: 152975688
AIM: mstover777

---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org


---------------------------------
Do you Yahoo!?
Yahoo! SiteBuilder - Free, easy-to-use web site design software