Diffs between pub/sub test in windows and linux

[Kuppe <ku...@360t.com>]

I am running a simple command-line driven pub/sub test where the broker is in
the same VM as the publisher and subscriber, just in different threads.
Surprisingly, when i run the test on a windows notebook, i get pretty good
results:

 1000 messages of 10 bytes each in 1051ms

but when i run the exact same test on a linux box which is much more
powerful than the windows notebook, i get completely different results:

 1000 messages of 10 bytes each in 80696ms

Can you suggest what could be different between the two environments that
could cause such variances?

Thanks!
--
View this message in context: http://www.nabble.com/Diffs-between-pub-sub-test-in-windows-and-linux-t1467236.html#a3965410
Sent from the ActiveMQ - User forum at Nabble.com.

Re: Diffs between pub/sub test in windows and linux

[Kuppe <ku...@360t.com>]

This is exactly as i would have expected!! I am using the 4.0 RC2 version...
Can you suggest a method of further isolating where i should be looking?

Perhaps somehow related is my previous posting regarding your proposed
configuration for a market data use case described on your web site...
Perhaps if i had a reference implementation i may be able to find the
problem myself (i would appreciate the configuration of both the broker and
the clients...)

Thanks in advance...


--
View this message in context: http://www.nabble.com/Diffs-between-pub-sub-test-in-windows-and-linux-t1467236.html#a3966986
Sent from the ActiveMQ - User forum at Nabble.com.

Re: Diffs between pub/sub test in windows and linux

[James Strachan <ja...@gmail.com>]

That does sound very strange. Are you using the vm:// transport or
tcp? If the latter it sounds like somethings seriously wrong with the
TCP stack as typically on unix tcp on the local box is super fast.

On 4/18/06, Kuppe <ku...@360t.com> wrote:
>
> I am running a simple command-line driven pub/sub test where the broker is in
> the same VM as the publisher and subscriber, just in different threads.
> Surprisingly, when i run the test on a windows notebook, i get pretty good
> results:
>
>  1000 messages of 10 bytes each in 1051ms
>
> but when i run the exact same test on a linux box which is much more
> powerful than the windows notebook, i get completely different results:
>
>  1000 messages of 10 bytes each in 80696ms
>
> Can you suggest what could be different between the two environments that
> could cause such variances?
>
> Thanks!
> --
> View this message in context: http://www.nabble.com/Diffs-between-pub-sub-test-in-windows-and-linux-t1467236.html#a3965410
> Sent from the ActiveMQ - User forum at Nabble.com.
>
>


--

James
-------
http://radio.weblogs.com/0112098/