Performance metrics and commercial deployment info required [message #78181] |
Wed, 29 November 2006 14:40 ![Go to next message Go to next message](theme/Solstice/images/down.png) |
Eclipse User![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=) |
|
|
|
Originally posted by: shaneosullivan1.gmail.com
Hi,
My company is considering using Equinox as a solution for a pluggable web
architecture. However first we need to know if there are any major
bottlenecks in performance or limitations on particular platforms. Does
anyone know of and performance metrics that have been performed against
Equinox, especially in comparison to other web archetectures?
Also, are there any web platforms that this is known to work on? Most posts
to this list mention using either Tomcat or Jetty - has anyone else used
WebSphere or WebLogic? If so, were any issues encountered that are
particular to that system?
Does anyone know of any commercial applications that either have been
deployed using Equinox, or are in development?
Thanks,
Shane O'Sullivan
|
|
|
Re: Performance metrics and commercial deployment info required [message #78275 is a reply to message #78181] |
Fri, 01 December 2006 05:01 ![Go to previous message Go to previous message](theme/Solstice/images/up.png) ![Go to next message Go to next message](theme/Solstice/images/down.png) |
Eclipse User![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=) |
|
|
|
Originally posted by: jeff_mcaffer.REMOVE.ca.ibm.com
You don't say exactly which parts of Equinox you are considering. In
any event, I don't know of any particular performance metrics or
benchmarks that have been run.
For the basic framework function you may not be aware that Websphere
(the new version) is actually based on Equinox and BEA is basing their
new microkernel server architecture (not sure the right name) on it as
well. There likely are still places that could be improved of course,
but things seem to be working pretty well.
For other parts of Equinox it will really depend. Some bundles are
extremely highly optimized (registry) while others are more classic
reference implementation (DS). Others are largely delegation wrappers
for other things (e.g., the server side stuff just wrappers Jetty or
forwards through requests being handled from the underlying server
infrastructure). It would be interesting actaully to get some numbers
where possible.
Suggestions? Volunteers?
Jeff
Shane O'Sullivan wrote:
> Hi,
>
> My company is considering using Equinox as a solution for a pluggable web
> architecture. However first we need to know if there are any major
> bottlenecks in performance or limitations on particular platforms. Does
> anyone know of and performance metrics that have been performed against
> Equinox, especially in comparison to other web archetectures?
>
> Also, are there any web platforms that this is known to work on? Most posts
> to this list mention using either Tomcat or Jetty - has anyone else used
> WebSphere or WebLogic? If so, were any issues encountered that are
> particular to that system?
>
> Does anyone know of any commercial applications that either have been
> deployed using Equinox, or are in development?
>
> Thanks,
>
> Shane O'Sullivan
>
>
|
|
|
Re: Performance metrics and commercial deployment info required [message #78350 is a reply to message #78275] |
Sun, 03 December 2006 05:36 ![Go to previous message Go to previous message](theme/Solstice/images/up.png) |
Eclipse User![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=) |
|
|
|
Originally posted by: skaegi.sympatico.ca
"Jeff McAffer" wrote...
> (e.g., the server side stuff just wrappers Jetty or
> forwards through requests being handled from the underlying server
> infrastructure). It would be interesting actaully to get some numbers
> where possible.
Yep. This is on the TODO list so I guess I'll volunteer ;)
I did some very basic load testing of the servletbridge a bit more than a
year ago and found that the associated overhead for "servlet" requests was
not noticeable even for incredibly simple servlets. Repeated "resource"
requests on the otherhand were slower since at the time there was no support
for the various HTTP cache control headers however support has since been
added. Most appservers support an in memory resource cache which "might" be
a further optimization worth looking into.
A good question is perhaps what we could use to test. Does anyone know of
any good standard benchmark suites we might use?
> Shane O'Sullivan wrote:
> > Hi,
> >
> > My company is considering using Equinox as a solution for a pluggable
web
> > architecture. However first we need to know if there are any major
> > bottlenecks in performance or limitations on particular platforms. Does
> > anyone know of and performance metrics that have been performed against
> > Equinox, especially in comparison to other web archetectures?
> >
I can offer one other anecdotal comment. Compared to large web applications
(say involving >100 jars) OSGi based web applications seem to initialize
noticeably more quickly. I suspect the classloading in OSGi based
applications is better able to target the correct place to do class lookups
but haven't done much analysis here.
> > Also, are there any web platforms that this is known to work on? Most
posts
> > to this list mention using either Tomcat or Jetty - has anyone else used
> > WebSphere or WebLogic? If so, were any issues encountered that are
> > particular to that system?
As far as I know the servletbridge should work on all app servers that
support servlet 2.3 and later. If an app server doesn't work then a bug
should be logged.
> >
> > Does anyone know of any commercial applications that either have been
> > deployed using Equinox, or are in development?
See Jeff's comments in the previous post. There are several groups doing
commercial development using server-side Eclipse.
|
|
|
Powered by
FUDForum. Page generated in 0.01819 seconds