You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@tomcat.apache.org by Shlomi Tsur <sh...@gmail.com> on 2008/01/02 11:34:33 UTC
Tomcat 6's nio protocol is slower than the regular HTTP/1.1
protocol
Hi
We are testing the new nio connector, and in our load tests we are getting
30% better results
in terms of fetches/sec in favor of the old HTTP/1.1 protocol.
All we are doing in the servlet is just returning status 200 in the
response, nothing more.
Does the new io really slower then the regular protocol?
Thank you.
--
View this message in context: http://www.nabble.com/Tomcat-6%27s-nio-protocol-is-slower-than-the-regular-HTTP-1.1-protocol-tp14576203p14576203.html
Sent from the Tomcat - User mailing list archive at Nabble.com.
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Tomcat 6's nio protocol is slower than the regular HTTP/1.1 protocol
Posted by Filip Hanik - Dev Lists <de...@hanik.com>.
that's expected, maybe not 30% but it all depends on the test.
What the NIO allows you for, is to have more sockets than threads, and
also being able to free up a worker thread when sending down static
content to the client.
Filip
Shlomi Tsur wrote:
> Hi
>
> We are testing the new nio connector, and in our load tests we are getting
> 30% better results
>
> in terms of fetches/sec in favor of the old HTTP/1.1 protocol.
>
> All we are doing in the servlet is just returning status 200 in the
> response, nothing more.
>
> Does the new io really slower then the regular protocol?
>
> Thank you.
>
>
>
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Tomcat 6's nio protocol is slower than the regular
HTTP/1.1protocol
Posted by David Cassidy <da...@twocats.co.uk>.
Did anyone do the comparison between ajp13 and http for the protocol ?
I'd like to understand what the test cases were that were used for the
test.
ie if the application takes a second to make the resulting html
if it takes 2 seconds to make the html how does that affect the
performance ?
D
On Wed, 2008-01-02 at 17:02 +0100, Rainer Jung wrote:
> Caldarale, Charles R schrieb:
> >> Does the new io really slower then the regular protocol?
> >
> > Yes; read this:
> > http://marc.info/?l=tomcat-user&m=119635696312254&w=2
> >
> > - Chuck
>
> and: speed != performance
>
> As an approximation:
>
> Throughput * AvgResponseTime = Parallelity
>
> Naively speed is the same as average response time. In most cases its
> better to think of performance as high throughput with acceptable
> response time.
>
> The formula shows you, that a solution, that is able to handle a very
> high parallelity could have a better performance (=throughput), than
> another one, even if the average response time might be worse.
>
> The NIO connector should be good in allowing a higher parallelism and
> thus achieve good throughput, even if the response times are worse than
> with the traditional connector.
>
> How important this is depends on the application scenario.
>
> Its never enough to only think about speed or only think about
> throughput when talking about high load. You need to define your minimal
> requirements for both of them and then search for the corresponding
> solution.
>
> Regards,
>
> Rainer
>
> ---------------------------------------------------------------------
> To start a new topic, e-mail: users@tomcat.apache.org
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Tomcat 6's nio protocol is slower than the regular HTTP/1.1protocol
Posted by Rainer Jung <ra...@kippdata.de>.
Caldarale, Charles R schrieb:
>> Does the new io really slower then the regular protocol?
>
> Yes; read this:
> http://marc.info/?l=tomcat-user&m=119635696312254&w=2
>
> - Chuck
and: speed != performance
As an approximation:
Throughput * AvgResponseTime = Parallelity
Naively speed is the same as average response time. In most cases its
better to think of performance as high throughput with acceptable
response time.
The formula shows you, that a solution, that is able to handle a very
high parallelity could have a better performance (=throughput), than
another one, even if the average response time might be worse.
The NIO connector should be good in allowing a higher parallelism and
thus achieve good throughput, even if the response times are worse than
with the traditional connector.
How important this is depends on the application scenario.
Its never enough to only think about speed or only think about
throughput when talking about high load. You need to define your minimal
requirements for both of them and then search for the corresponding
solution.
Regards,
Rainer
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
RE: Tomcat 6's nio protocol is slower than the regular HTTP/1.1protocol
Posted by "Caldarale, Charles R" <Ch...@unisys.com>.
> From: Shlomi Tsur [mailto:shlomo.tsur@gmail.com]
> Subject: Tomcat 6's nio protocol is slower than the regular
> HTTP/1.1protocol
>
> Does the new io really slower then the regular protocol?
Yes; read this:
http://marc.info/?l=tomcat-user&m=119635696312254&w=2
- Chuck
THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY
MATERIAL and is thus for use only by the intended recipient. If you
received this in error, please contact the sender and delete the e-mail
and its attachments from all computers.
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
RE: Tomcat 6's nio protocol is slower than the regular HTTP/1.1
protocol
Posted by Martin Gainty <mg...@hotmail.com>.
> > Hi> > We are testing the new nio connector, and in our load tests we are getting> 30% better results> > in terms of fetches/sec in favor of the old HTTP/1.1 protocol.> > All we are doing in the servlet is just returning status 200 in the> response, nothing more.> > Does the new io
Re>what new io?
really slower then the regular protocol?
Re>do you mean HTTP/1.0 ?> > Thank you.Re>You're welcome> > -- > View this message in context: http://www.nabble.com/Tomcat-6%27s-nio-protocol-is-slower-than-the-regular-HTTP-1.1-protocol-tp14576203p14576203.html> Sent from the Tomcat - User mailing list archive at Nabble.com.> > > ---------------------------------------------------------------------> To start a new topic, e-mail: users@tomcat.apache.org> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org> For additional commands, e-mail: users-help@tomcat.apache.org>
_________________________________________________________________
Share life as it happens with the new Windows Live.
http://www.windowslive.com/share.html?ocid=TXT_TAGHM_Wave2_sharelife_122007