Hi,
great news :)
Regarding 141k result you see, IMO it's related to Grizzly client
(specifically async write queue implementation), which is not optimized
for your usecase.
Going to fix it this asap.
Thanks.
WBR,
Alexey.
On 09/11/2011 12:48 PM, BlueDavy Lin wrote:
> hi!
>
> Because I'm not in grizzly dev maillist,so I have to write a new
> mail for this.
>
> I merged Oleksiy Stashok patch for nfs-rpc grizzly,and do the
> benchmark tests again (because I cann't get grizzly 2.2-snapshot,so I
> use grizzly 2.1.2 to do the test) ,the results is promising,thks.
>
> Now in 100 concurrents,100 request byte size,100 response byte
> size,single connection situation,the mina/netty/grizzly best results
> are created with protobuf,results are:
> Mina: 120k
> Netty: 153k
> Grizzly: 141k
>
> But in multi connections situation,such as two connections or
> three connections,grizzly is the best,when use three
> connections,grizzly create 190k record,it's amazing.
>
> More benchmark results pls see here: http://code.google.com/p/nfs-rpc/
>