On 06/03/07, Alexander J Turner <[EMAIL PROTECTED]> wrote:
Marlon,
This is really important to get clear. You cannot load test one machine
with one machine unless there is a large difference is performance
between the server (small) and client (large). This is especially true
of JMeter, which is not tuned for performance. A Apache linux server or
Windows IIS server will totally and utterly outperform JMeter without
breaking out a sweat.
That depends somewhat on how well the application is written - and
what it is doing.
JMeter running a simple test plan can generate quite a substantial
load; if the server has to do a lot of work for each request it may
well be possible to overload the server.
However, I agree that the machine running JMeter should not be running
at or near its limit, otherwise the test results may be somewhat
variable. Likewise the network connection from the test machine needs
to be able to cope with the load.
Further, if it is possible for one client
machine to load your server significantly, you have an issue!
Depends on what the server is designed to do ...
And may depend on the test plan. e.g. if the server provides different
services, some of which are rarely needed but which are very resource
intensive, then it may well be possible for a few client threads to
cause a problem for the server. But that does not necessarily mean the
server is faulty.
Please have this in mind when you are designing your test cases.
S.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
