Salut,
Oleksiy Stashok wrote:
> Hi Jeanfrancois,
>
> I don't have any bm results... But what I can propose to do first - is
> to change ReadFilter to try the read just once, instead of 3 times.
Ok we can ask. At least for JDK 1.4 and 1.5 it proved to be useful to
try more than one time.
And
> ask Scott to check the perf. numbers, if it made any difference... If
> not - we have to make setContinuousExecution(false) by default.
Agree. Scott, can I send you a patch to test :-)
A+
-- Jeanfrancois
>
> Thanks.
>
> WBR,
> Alexey.
>
> On Aug 26, 2008, at 0:06 , Jeanfrancois Arcand wrote:
>
>> Salut,
>>
>> we are seeing a performance improvement in GlassFish v3 when we set
>> the DefaultProtocolChain.setContinuousExecution(false). For the first
>> couple of Grizzly's release, that value was false as well, but we then
>> decided based on some testing to turn it to true. I don't recall who
>> was doing the testing (probably me), but it seems, at least for http,
>> that we are getting better result. Should we roll back to false by
>> default?
>>
>> https://grizzly.dev.java.net/issues/show_bug.cgi?id=57
>>
>> I just want to make sure the performance improvement is not only
>> applying to http. Can people using the DefaultProtocolChain and
>> ReadFilter gives it a try to see what kind of performance they are
>> getting when setting the value to false?
>>
>> A+
>>
>> -- Jeanfrancois
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe_at_grizzly.dev.java.net
>> For additional commands, e-mail: dev-help_at_grizzly.dev.java.net
>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe_at_grizzly.dev.java.net
> For additional commands, e-mail: dev-help_at_grizzly.dev.java.net
>