Thanks Rob,

netty.io is more independent and has way fewer and cleaner dependencies as
far as I can see. Therefore, I am leaning towards netty.io as well. Most
important though would be to pick one implementation and focus on that
implementation, not split effort between netty.io and undertow. As a
second requirement, the http stack should be light, I wouldn’t want to
carry around half a jboss in my http-undertow bundle :=)

Best
Frank

On 4/15/15, 1:54 PM, "Rob Walker" <[email protected]> wrote:

>Done a bit of background reading and both netty and undertow sound
>interesting - they both score pretty high on the performance graph
>linked to in the OP.
>
>Any particular reason to go undertow vs netty. Seems netty started in
>JBOSS but then move solo when they wanted a more to-the-metal approach
>which became undertow. Netty seems a bit more feature rich at this
>stage, which could perhaps be useful/interesting.
>
>Of course the beauty of OSGi and projects like Felix etc is that there's
>no reason not to have multiple HttpService implementations, assuming
>there are enough contributors.
>
>FWIW - our Felix project is on a very old cut of the Jetty based service
>and I have a background action to move to a newer one. I'm not sure I'll
>have time to help in implementing a netty/undertow based implementation
>(although I haven't ruled that out) - but could be a guinea pig in
>helping test it out in a real OSGi based App.
>
>-- Rob
>
>On 14/04/2015 15:47, Achim Nierbeck wrote:
>> Hi Nick,
>>
>> Right now the "backlog" is a bit thin [1], but help is certainly
>> appreciated :)
>> The main issue PAXWEB-771 still needs to collect the items to do.
>> But right now neither Harald or I do have a lot of free cycles for it :(
>> It might be better to move this discussion to the ops4j mailinglist,
>>that's
>> why I cross posted this answer.
>>
>> regards, Achim
>> [1] - https://ops4j1.jira.com/projects/PAXWEB/issues
>>
>> 2015-04-14 15:41 GMT+02:00 Nick Baker <[email protected]>:
>>
>>> We (Pentaho) can help with this. We¹d like to transition off Tomcat
>>> clusters to Karaf instances. One of the issues with this migration
>>>we¹ve
>>> identified is Jetty performance. Our next release will go out Tomcat
>>>with
>>> Karaf embedded, Felix HTTP Bridge spanning the two. Not where we want
>>>to
>>> be! What¹s the timeframe for the merge? Do you have a backlog of issues
>>> related to this? Let us know how we can help.
>>>
>>> -Nick Baker
>>>
>>> On 4/14/15, 7:31 AM, "Achim Nierbeck" <[email protected]> wrote:
>>>
>>>> Hi,
>>>>
>>>> nope it's actually Pax-Web 5.0 M1 and M2 (from the pax-web-undertow
>>>> branch)
>>>> which uses undertow only. And it's none of the fabric8 developers it's
>>>> Harald Wellmann, who also takes care of the Pax Exam.
>>>>
>>>> Right now I'm trying to find a way of having undertow as the third
>>>> supported container for pax web, to merge the pax-web-undertow
>>>> branch into the Pax-Web main branch for a 5.0
>>>> Help appreciated by the way ;)
>>>>
>>>> regards, Achim
>>>>
>>>>
>>>> 2015-04-14 13:28 GMT+02:00 Nick Baker <[email protected]>:
>>>>
>>>>> I could have swore one of the fabric8 developers was working on a
>>>>> pax-web-wildfly which is undertow-based. Not finding it via search.
>>>>>
>>>>> -Nick
>>>>>    Original Message
>>>>> From: Rob Walker
>>>>> Sent: Tuesday, April 14, 2015 5:44 AM
>>>>> To: [email protected]
>>>>> Reply To: [email protected]
>>>>> Subject: Re: Jetty vs Undertow
>>>>>
>>>>>
>>>>> Interesting. Jetty is quite a long way down that graph compared to
>>>>> those.
>>>>>
>>>>> I know Richard (founder of Oscar which became Felix) was always
>>>>> interested in light/fast HttpService implementations. Could be an
>>>>> interesting little side project.
>>>>>
>>>>> Wiring Jetty into the very original OSGi HttpService interface didn't
>>>>> actually take that much work, although I haven't been back to the
>>>>>spec
>>>>> of late to see what's changed or been added that might make things
>>>>> harder
>>>>>
>>>>> -- Rob
>>>>>
>>>>> On 14/04/2015 11:27, Frank Langel wrote:
>>>>>> When using Felix http module in performance critical applications,
>>>>> Jetty
>>>>>> seems to underperform compared to modern servlet libraries such as
>>>>> undertow
>>>>>> or netty:
>>>>>>
>>>>>
>>> 
>>>https://www.techempower.com/benchmarks/#section=data-r9&hw=peak&test=jso
>>>n
>>>>>> Has/Is anyone working on a plug-in replacement for Undertow/Netty.
>>>>>>Any
>>>>>> experience/performance measurements? Any plans?
>>>>>>
>>>>>> Thanks a lot in advance
>>>>>> Frank
>>>>>>
>>>>>>
>>>>>>
>>>>> --
>>>>>
>>>>>
>>>>> Ascert - Taking systems to the edge
>>>>> [email protected]
>>>>> SA +27 21 300 2028
>>>>> UK +44 20 7488 3470 ext 5119
>>>>> www.ascert.com
>>>>>
>>>>>
>>>>> ---------------------------------------------------------------------
>>>>> To unsubscribe, e-mail: [email protected]
>>>>> For additional commands, e-mail: [email protected]
>>>>>
>>>>>
>>>>> ---------------------------------------------------------------------
>>>>> To unsubscribe, e-mail: [email protected]
>>>>> For additional commands, e-mail: [email protected]
>>>>>
>>>>>
>>>>
>>>> --
>>>>
>>>> Apache Member
>>>> Apache Karaf <http://karaf.apache.org/> Committer & PMC
>>>> OPS4J Pax Web <http://wiki.ops4j.org/display/paxweb/Pax+Web/>
>>>>Committer &
>>>> Project Lead
>>>> blog <http://notizblog.nierbeck.de/>
>>>> Co-Author of Apache Karaf Cookbook <http://bit.ly/1ps9rkS>
>>>>
>>>> Software Architect / Project Manager / Scrum Master
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: [email protected]
>>> For additional commands, e-mail: [email protected]
>>>
>>>
>>
>
>-- 
>
>
>Ascert - Taking systems to the edge
>[email protected]
>SA +27 21 300 2028
>UK +44 20 7488 3470 ext 5119
>www.ascert.com
>
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: [email protected]
>For additional commands, e-mail: [email protected]
>



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to