users@grizzly.java.net

Re: performance comparison: nio v nio2

From: Jeanfrancois Arcand <Jeanfrancois.Arcand_at_Sun.COM>
Date: Mon, 08 Jun 2009 12:42:09 -0400

Salut,

John Franey wrote:
> I'm wondering if anyone can share measured performance comparisons
> between selector and async io techniques.
>
> In my mind, I can imagine that async io has better throughput. I will
> measure difference for my own sake, but what can I expect?

yes, I did. So far all the test proved that AIO was faster. I didn't
publish my results last week during my session @ JavaOne, but I do see a
~15% improvement in some case. For me I've mainly tested the http
modules's, and yes throughput was better. Let me catch up with the
Grizzly issues this week but I can share some of the test I did.

Thanks!

-- Jeanfrancois


>
> Thanks,
> John
>