On Tue, Sep 14, 2004 at 09:55:41PM +0300, [EMAIL PROTECTED] wrote:
> >It's possible, of course, that the nature of the task at hand is such
> >that it's impossible to satisfy condition a), because the code that does
> >the actual job entrusted to the server *must* wait for some resources.
>
> Or d
Micha Feigin wrote:
On Tue, Sep 14, 2004 at 09:35:13AM +0300, [EMAIL PROTECTED] wrote:
Daniel Feiglin wrote:
Is this some kind of thread per client application?
If it is, you might like to take a look at the good old C library
select(2) api which has been somehow implemented in Java 1.4.x (see for
Anatoly Vorobey wrote:
[EMAIL PROTECTED] wrote on Tue, Sep 14, 2004 at 09:35:13AM:
If it is, you might like to take a look at the good old C library
select(2) api which has been somehow implemented in Java 1.4.x (see for
example, API doc java.nio.channels.SelectableChannel). I haven't used
this Ja
On Tue, Sep 14, 2004 at 12:16:37PM +0300, Eli Kara wrote:
> > I wrote (and have been maintaining) a network memory cache server using
> > the single-thread model and libevent. It routinely handles many hundreds
> > of simultaneous clients, easily scales to over a thousand, and does all
> > this usi
On Tue, Sep 14, 2004 at 09:35:13AM +0300, [EMAIL PROTECTED] wrote:
> Daniel Feiglin wrote:
>
> >Is this some kind of thread per client application?
> >
> >If it is, you might like to take a look at the good old C library
> >select(2) api which has been somehow implemented in Java 1.4.x (see for
>
On Tuesday 14 September 2004 10:44, [EMAIL PROTECTED] wrote:
> Oded Arbel wrote:
> >On Tuesday 14 September 2004 09:35, [EMAIL PROTECTED] wrote:
> >>Now you say that the benefit of using select is to cut-down the
> >>number of threads/processes -
> >>what do you expect to gain from this? Reduce mem
On Tue, Sep 14, 2004 at 09:50:08AM +0200, [EMAIL PROTECTED] wrote:
> I can not say anything about the performance of 1:1 linux threading model.
> I can not say that it is supperior or anything like that. I have not
> done any benchmarks. I can just say that the design of this model has
> some draw
On Tuesday 14 September 2004 10:45, Omer Zak wrote:
> One thing which I do not understand:
>
> Why use Java at all in a system, which needs to handle hundreds or
> thousands of concurrent HTTP calls, and therefore can justify the
> effort to be invested in optimizing thread handling by using C/C++
>
> I wrote (and have been maintaining) a network memory cache server using
> the single-thread model and libevent. It routinely handles many hundreds
> of simultaneous clients, easily scales to over a thousand, and does all
> this using very little CPU (never more than 1-2%). It'd be nice to be
>
Quoting Muli Ben-Yehuda <[EMAIL PROTECTED]>:
I can not say anything about the performance of 1:1 linux threading model.
I can not say that it is supperior or anything like that. I have not
done any benchmarks. I can just say that the design of this model has
some drawbacks.
"However, the two majo
Oded Arbel wrote:
On Tuesday 14 September 2004 09:35, [EMAIL PROTECTED] wrote:
Now you say that the benefit of using select is to cut-down the
number of threads/processes -
what do you expect to gain from this? Reduce memory use?
Reduced overhead. The overhead of context switching between m
One thing which I do not understand:
Why use Java at all in a system, which needs to handle hundreds or
thousands of concurrent HTTP calls, and therefore can justify the effort
to be invested in optimizing thread handling by using C/C++ implementation?
Oded Arbel wrote:
On Tuesday 14 September 2
[EMAIL PROTECTED] wrote on Tue, Sep 14, 2004 at 09:35:13AM:
> >If it is, you might like to take a look at the good old C library
> >select(2) api which has been somehow implemented in Java 1.4.x (see for
> >example, API doc java.nio.channels.SelectableChannel). I haven't used
> >this Java API myse
On Tuesday 14 September 2004 09:35, [EMAIL PROTECTED] wrote:
> Now you say that the benefit of using select is to cut-down the
> number of threads/processes -
> what do you expect to gain from this? Reduce memory use?
Reduced overhead. The overhead of context switching between multiple
threads, e
Daniel Feiglin wrote:
Is this some kind of thread per client application?
If it is, you might like to take a look at the good old C library
select(2) api which has been somehow implemented in Java 1.4.x (see for
example, API doc java.nio.channels.SelectableChannel). I haven't used
this Java API mys
On Mon, Sep 13, 2004 at 08:08:29PM +0200, [EMAIL PROTECTED] wrote:
> So it is better to switch to NetBSD.
I'll assume for a second that you aren't just trolling. Care to
elaborate on why you think M:N is superior, considering the details
and performance of the Linux implementation of 1:1?
Cheers
On Mon, 13 Sep 2004, Daniel Feiglin wrote:
> Is this some kind of thread per client application?
>
> If it is, you might like to take a look at the good old C library
> select(2) api which has been somehow implemented in Java 1.4.x (see for
> example, API doc java.nio.channels.SelectableChannel).
> >
> > U better try to test the new kernel version 2.6.x.
> > It has support for userspace semaphores
>
> They're called 'futexes'.
>
> > and support
> > for mapping number of kernel threads for different number
> > of userspace threads. I think it is called N:M model.
>
> It's called 'M:N thread
On Mon, Sep 13, 2004 at 10:20:40AM +0200, [EMAIL PROTECTED] wrote:
>
> Hi Vladek
>
> U better try to test the new kernel version 2.6.x.
> It has support for userspace semaphores
They're called 'futexes'.
> and support
> for mapping number of kernel threads for different number
> of userspace
On Monday 13 September 2004 10:46, Sobolevsky Vladek wrote:
> Hi All,
> I have problem to run more then 1000 threads on Linux.
> ...
> I am running Red Hat 7.2 , 2.4.18 kernel.
Even if/when it works, it would create huge scalability problems
due to the Linux threading model of this old distro.
Fo
Is this some kind of thread per client application?
If it is, you might like to take a look at the good old C library
select(2) api which has been somehow implemented in Java 1.4.x (see for
example, API doc java.nio.channels.SelectableChannel). I haven't used
this Java API myself, but if it's anyth
On Monday 13 September 2004 11:10, Eli Kara wrote:
> On Monday 13 September 2004 10:46, Sobolevsky Vladek wrote:
> > Hi All,
> > I have problem to run more then 1000 threads on Linux.
> > I get Out of Memory error, the program written in Java so I suspect
> > it related
> Have you tried increasing
Hi Vladek
U better try to test the new kernel version 2.6.x.
It has support for userspace semaphores and support
for mapping number of kernel threads for different number
of userspace threads. I think it is called N:M model.
In general it can give you some imprevements on the
same hardware.
Also
On Monday 13 September 2004 10:46, Sobolevsky Vladek wrote:
> Hi All,
> I have problem to run more then 1000 threads on Linux.
> I get Out of Memory error, the program written in Java so I suspect it
> related
> To Java thread implementation. This limitation applies only to one
> process , I can ru
Hi All,
I have problem to run more then 1000 threads on Linux.
I get Out of Memory error, the program written in Java so I suspect it
related
To Java thread implementation. This limitation applies only to one
process , I can run
For example 2 processes that will use 900 threads each and it works
f
25 matches
Mail list logo
