Hi,
I did also create a version of SocketIoProcessor for use with aio4j. However it proved to be unreliable under linux (at least RHEL 4) although it was ok on windows. AIO will be part of Java 7 apparently so I will hopefully revisit it when that is available in early access.
Sun are currently trying to reproduce my test environment in their lab, so I should stress that this is not a validated bug or problem in Mustang at this stage. I will keep this mailing list updated with progress on this issue - it could be some obscure characteristic of my test.
I am looking forward to BEA's version of Java 6 though! In other testing we have done, we found that Mustang did show significant improvements in the cost of acquiring locks - i.e. sychronized blocks etc - presumably due to the escape analysis and other optimisations that are done in that area.
Robert
| "Vinod Panicker" <[EMAIL PROTECTED]>
18/04/2006 14:07
|
To: [email protected] cc: Subject: Re: MINA with Mustang |
Hi,
Thanks for the response. This is eye-opening info...
Have you ever tried IBM's aio4j?
Anyways... I hope sun manages to get the performance atleast up to
jrockit standards when Mustang goes final.
Regards,
Vinod.
On 4/18/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
>
>
> Yes, we have been using Mustang betas for quite some time with our application.
>
> Our results show that, on our HP DL 585 hardware (4 dual core opterons) BEA JRockit is still faster in terms of throughput than the Sun JVM. Most interestingly, the BEA JVM is significantly faster when Gb ethernet is connecting the clients and servers. We are running RHEL 4.
>
> I am working with Sun at the moment to analyse our performance results to see if we can figure out why we see such a difference between the JVMs particularly when using Gb ethernet.
>
> Our benchmarks involve sending payloads of around 400 bytes from clients to server. We have tested with up to 150 clients running concurrently.
>
> Robert
>
>
>
>
> "Vinod Panicker" <[EMAIL PROTECTED]>
>
> 18/04/2006 11:32
> Please respond to mina-dev
> To: [email protected]
> cc:
> Subject: MINA with Mustang
>
>
>
> Hi,
>
> Has anyone given MINA a try with the new JDK 1.6 beta?
>
> It's got the epoll based selector implementation and should be much
> more efficient in managing more number of concurrent connections. CPU
> usage with idle connections should ideally drop drastically.
>
> If anyone's got the time and hardware, some benchmarks would be really
> welcome :)
>
> Regards,
> Vinod.
This communication is for informational purposes only. It is not intended
as an offer or solicitation for the purchase or sale of any financial
instrument or as an official confirmation of any transaction. All market prices,
data and other information are not warranted as to completeness or accuracy and
are subject to change without notice. Any comments or statements made herein
do not necessarily reflect those of JPMorgan Chase & Co., its subsidiaries
and affiliates.
