Re: [jetty-dev] NIO inefficiency?

[Simone Bordet <sbordet@xxxxxxxxxxx>]

Hi,

On Fri, Aug 9, 2013 at 10:38 AM, Viktor Szathmary <phraktle@xxxxxxxxx> wrote:
> Hi Simone,
>
> On Friday, August 9, 2013 at 10:03 AM, Simone Bordet wrote:
>
> What profiler and what mode (sampling, instrumentation, etc.) ?
>
> This hotspot was corroborated by two different profilers (New Relic and
> YourKit). Both of these were sampling profilers,  used with relatively low
> sampling frequencies over longer time periods. I believe this could have
> inaccuracies due to stack traces being taken only on JVM safe-points – but
> assuming it's not an artifact, what would make for a good experiment to
> prove/disprove this theory?

That's a good question, and I don't have a clear answer.
In the past, I used different profilers where one was showing an
artifact but the other did not.

Sampling is indeed affected by JVM safe-pointing (mostly
time-to-safe-point) in almost any profiler.
There seems to be a new technique using AsyncGetCallTrace() (see
http://jeremymanson.blogspot.it/2013/07/lightweight-asynchronous-sampling.html)
which is not subject to safe-pointing problems, but AFAIK no current
commercial profiler uses it, apart probably Sun Studio Analyzer (aka
Oracle Solaris Studio) - which works on Linux too.

If you try that, let us know.

-- 
Simone Bordet
----
http://cometd.org
http://webtide.com
http://intalio.com
Developer advice, training, services and support
from the Jetty & CometD experts.
Intalio, the modern way to build business applications.