NIO vs NPTL+standard IO

[Stefano Bagnara <ap...@bago.org>]

Today I found another blog post on the "usual" topic:
http://mailinator.blogspot.com/2008/02/kill-myth-please-nio-is-not-faster-than.html

Is this FUD or a new trend that MINA should take care of?

Is there any plan to support standard IO as a transport for MINA based 
applications?

It would be useful to understand how a MINA based application could 
"switch back" to multithreaded standard IO and result in better 
throughput on modern JVM/OS.

Thank you,
Stefano

Re: NIO vs NPTL+standard IO

[Julien Vermillard <jv...@archean.fr>]

On Tue, 19 Feb 2008 14:20:59 +0100
Niklas Therning <ni...@trillian.se> wrote:

> David M. Lloyd wrote:
> > On 02/19/2008 07:07 AM, Stefano Bagnara wrote:
> >> Today I found another blog post on the "usual" topic:
> >> http://mailinator.blogspot.com/2008/02/kill-myth-please-nio-is-not-faster-than.html 
> >>
> >> Is this FUD or a new trend that MINA should take care of?
> >
> > This test was run on a very old kernel, with an old JVM.  I'm not
> > too worried. :-)
> >
> > - DML
> >
> Yes, I agree, the original test is almost 4 years old and uses Java 
> 1.4.2 with an old Linux kernel. But it would still be interesting to
> see the results when using a modern kernel and Java 6. Has anyone
> seen such a test?
> 
> /Niklas
> 

NIO now use epoll so this test is really deprecated.

Anyway if someone make a good perf test suite, I'll take a look for
measuring the real gap between NIOTransport and the APR one :)

Julien

Re: NIO vs NPTL+standard IO

[Niklas Therning <ni...@trillian.se>]

David M. Lloyd wrote:
> On 02/19/2008 07:07 AM, Stefano Bagnara wrote:
>> Today I found another blog post on the "usual" topic:
>> http://mailinator.blogspot.com/2008/02/kill-myth-please-nio-is-not-faster-than.html 
>>
>> Is this FUD or a new trend that MINA should take care of?
>
> This test was run on a very old kernel, with an old JVM.  I'm not too 
> worried. :-)
>
> - DML
>
Yes, I agree, the original test is almost 4 years old and uses Java 
1.4.2 with an old Linux kernel. But it would still be interesting to see 
the results when using a modern kernel and Java 6. Has anyone seen such 
a test?

/Niklas

Re: NIO vs NPTL+standard IO

[Mark Webb <el...@gmail.com>]

Based on Paul Tyma's background and track record just with the
mailinator program, I would not say that he is some sort of hack just
looking to get a rise out of people.

I do think it would be interesting to write a mailinator-type system
using MINA to see how well it could perform....



On Feb 19, 2008 8:17 AM, David M. Lloyd <da...@redhat.com> wrote:
> Haha, and the blog owner is moderating comments.  I'm sure the truth will
> not see the light of day for this guy.
>
>
> On 02/19/2008 07:13 AM, David M. Lloyd wrote:
> > On 02/19/2008 07:07 AM, Stefano Bagnara wrote:
> >> Today I found another blog post on the "usual" topic:
> >> http://mailinator.blogspot.com/2008/02/kill-myth-please-nio-is-not-faster-than.html
> >>
> >> Is this FUD or a new trend that MINA should take care of?
> >
> > This test was run on a very old kernel, with an old JVM.  I'm not too
> > worried. :-)
> >
> > - DML
> >
>



-- 
--------------------------------
Talent hits a target no one else can hit; Genius hits a target no one
else can see.

Re: NIO vs NPTL+standard IO

["David M. Lloyd" <da...@redhat.com>]

Haha, and the blog owner is moderating comments.  I'm sure the truth will 
not see the light of day for this guy.

On 02/19/2008 07:13 AM, David M. Lloyd wrote:
> On 02/19/2008 07:07 AM, Stefano Bagnara wrote:
>> Today I found another blog post on the "usual" topic:
>> http://mailinator.blogspot.com/2008/02/kill-myth-please-nio-is-not-faster-than.html 
>>
>> Is this FUD or a new trend that MINA should take care of?
> 
> This test was run on a very old kernel, with an old JVM.  I'm not too 
> worried. :-)
> 
> - DML
> 

Re: NIO vs NPTL+standard IO

["David M. Lloyd" <da...@redhat.com>]

On 02/19/2008 07:07 AM, Stefano Bagnara wrote:
> Today I found another blog post on the "usual" topic:
> http://mailinator.blogspot.com/2008/02/kill-myth-please-nio-is-not-faster-than.html 
> 
> Is this FUD or a new trend that MINA should take care of?

This test was run on a very old kernel, with an old JVM.  I'm not too 
worried. :-)

- DML

Re: NIO vs NPTL+standard IO

[Mike Heath <mh...@apache.org>]

I've suspected for a while that you can getter better performance using
thread-per-connection on Java.  I've never actually tested this theory
though.  There are certainly a lot of scenarios where you can get lower
latency using thread-per-connection because you don't have the overhead
of queuing.

Perhaps we should start looking seriously at implementing
thread-per-connection IoServices for MINA so we can do some apples to
apples performance comparisons between NIO, IO, and APR.

-Mike

Stefano Bagnara wrote:
> Today I found another blog post on the "usual" topic:
> http://mailinator.blogspot.com/2008/02/kill-myth-please-nio-is-not-faster-than.html
> 
> 
> Is this FUD or a new trend that MINA should take care of?
> 
> Is there any plan to support standard IO as a transport for MINA based
> applications?
> 
> It would be useful to understand how a MINA based application could
> "switch back" to multithreaded standard IO and result in better
> throughput on modern JVM/OS.
> 
> Thank you,
> Stefano
> 
>