ServiceMix Websphere deployment performance

[shen <gs...@infocomp.com>]

Hi Guys,
Has anyone had experience with ServiceMix performance relate to the
Websphere deployment?

I inherited this project from previous developers, which deploys the
ServiceMix into Websphere App Server. 
Now our client has a performance requirement. When I tested the performance
on WAS, for the same request, the average response time is about 2-3 times
of running ServiceMix standalone.
e.g. same request, same throughput, 
when run ServiceMix standalone, average response time about 170ms (under
client's requirement)
when run in WAS, average about 500ms.(twice the time of the client's
requirement)

When using java projfiler tool, couldn’t see any significant difference in
percentage way. 
I noticed that the GC seems busier when run in WAS, more than 10% cpu, while
1-2% when run standalone.

let me know if you have any idea or suggestions.

Thanks,
George


-- 
View this message in context: http://www.nabble.com/ServiceMix-Websphere-deployment-performance-tf3788182s12049.html#a10712900
Sent from the ServiceMix - User mailing list archive at Nabble.com.

Re: ServiceMix Websphere deployment performance

["Bob G." <ae...@hotmail.com>]

Hello,

Can I ask you what is your version of WAS?

We have Websphere App Server 6.1. and we are evaluating the possibilities to
deploy ServiceMix into this container. 

Do you have some feed back on this type of deployment?





shen wrote:
> 
> Hi Guys,
> Has anyone had experience with ServiceMix performance relate to the
> Websphere deployment?
> 
> I inherited this project from previous developers, which deploys the
> ServiceMix into Websphere App Server. 
> Now our client has a performance requirement. When I tested the
> performance on WAS, for the same request, the average response time is
> about 2-3 times of running ServiceMix standalone.
> e.g. same request, same throughput, 
> when run ServiceMix standalone, average response time about 170ms (under
> client's requirement)
> when run in WAS, average about 500ms.(twice the time of the client's
> requirement)
> 
> When using java projfiler tool, couldn’t see any significant difference in
> percentage way. 
> I noticed that the GC seems busier when run in WAS, more than 10% cpu,
> while 1-2% when run standalone.
> 
> let me know if you have any idea or suggestions.
> 
> Thanks,
> George
> 
> 
> 

-- 
View this message in context: http://www.nabble.com/ServiceMix-Websphere-deployment-performance-tf3788182s12049.html#a11744343
Sent from the ServiceMix - User mailing list archive at Nabble.com.