You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@servicemix.apache.org by Guillaume Nodet <gn...@gmail.com> on 2010/05/17 12:39:27 UTC
[DISCUSS] ServiceMix RoadMap
Here are a few thoughts I'd like to throw in for the next version of
ServiceMix:
* documentation (we really, really need that)
* better support for Blueprint: we should support blueprint for everyhting
we have, it would include Camel, ActiveMQ and ServiceMix components
* better scalability:
- we need to put back the asynchronous support for Camel (in camel JBI
and NMR transports)
- we need to improve CXF-BC to better support streaming (the
useJbiWrapper on the consumer could still use streaming as done in
servicemix-http soap endpoints)
- we need to find a real nice http client to have non blocking threads
and where we would be able to write to the stream instead of having to give
it an input stream
I'll try to improve the doc on karaf and move it into a docbook format for
easier integration into the servicemix doc, i think that would help too.
I'll work on the Blueprint stuff, as I've already started.
Claus is putting back async support in Camel, so once it's done, we should
be able to put back the change we reverted for upgrading to 2.0 in both JBI
and NMR compoennts for camel.
CXF-BC I think still needs some work from a scalability / performance point
of view. Having an HTTP client that could really be leveraged would be
awesome, but I don't know any that would really work yet. I think
httpcomponents may be a good candidate as it should be able to be used in
asynchronous and streaming mode.
Also, I think we should try to do some perfs / load testing and have the
defaults config behave nicely, even if we need to break some compatibility
at some point. For example, having the cxf-bc consumer being asynchronous
by default would make sense to me.
--
Cheers,
Guillaume Nodet
------------------------
Blog: http://gnodet.blogspot.com/
------------------------
Open Source SOA
http://fusesource.com
Re: [DISCUSS] ServiceMix RoadMap
Posted by "Jamie G." <ja...@gmail.com>.
+1 to the above suggestions as well.
It would be good if a few JIRAs could be logged for the proposed tasks
so we can pick them up.
As to the benchmarking work, a perf-report feature would be neat. The
user would install the feature, and it would in turn load a few common
endpoints and execute a number of exchanges, capturing some perf
metrics (via the exchange listener?), then produce a summary report.
The idea is to make something easy to use so that users could easily
run the test as they tweak service settings.
On Tue, May 18, 2010 at 4:08 AM, Lars Heinemann <lh...@apache.org> wrote:
> Sounds all great, so +1 to all of the made proposals.
> Additionally I would like to push the cookbook recipes at our website a bit.
> It would be nice to reach a state like the Apache Camel recipes
> collection or even better ;)
>
> Regards
> Lars
>
>
>
> 2010/5/17 Guillaume Nodet <gn...@gmail.com>:
>> Here are a few thoughts I'd like to throw in for the next version of
>> ServiceMix:
>> * documentation (we really, really need that)
>> * better support for Blueprint: we should support blueprint for everyhting
>> we have, it would include Camel, ActiveMQ and ServiceMix components
>> * better scalability:
>> - we need to put back the asynchronous support for Camel (in camel JBI
>> and NMR transports)
>> - we need to improve CXF-BC to better support streaming (the
>> useJbiWrapper on the consumer could still use streaming as done in
>> servicemix-http soap endpoints)
>> - we need to find a real nice http client to have non blocking threads
>> and where we would be able to write to the stream instead of having to give
>> it an input stream
>>
>> I'll try to improve the doc on karaf and move it into a docbook format for
>> easier integration into the servicemix doc, i think that would help too.
>> I'll work on the Blueprint stuff, as I've already started.
>> Claus is putting back async support in Camel, so once it's done, we should
>> be able to put back the change we reverted for upgrading to 2.0 in both JBI
>> and NMR compoennts for camel.
>> CXF-BC I think still needs some work from a scalability / performance point
>> of view. Having an HTTP client that could really be leveraged would be
>> awesome, but I don't know any that would really work yet. I think
>> httpcomponents may be a good candidate as it should be able to be used in
>> asynchronous and streaming mode.
>>
>> Also, I think we should try to do some perfs / load testing and have the
>> defaults config behave nicely, even if we need to break some compatibility
>> at some point. For example, having the cxf-bc consumer being asynchronous
>> by default would make sense to me.
>>
>> --
>> Cheers,
>> Guillaume Nodet
>> ------------------------
>> Blog: http://gnodet.blogspot.com/
>> ------------------------
>> Open Source SOA
>> http://fusesource.com
>>
>
>
>
> --
> http://lhein.blogspot.com
>
Re: [DISCUSS] ServiceMix RoadMap
Posted by Lars Heinemann <lh...@apache.org>.
Sounds all great, so +1 to all of the made proposals.
Additionally I would like to push the cookbook recipes at our website a bit.
It would be nice to reach a state like the Apache Camel recipes
collection or even better ;)
Regards
Lars
2010/5/17 Guillaume Nodet <gn...@gmail.com>:
> Here are a few thoughts I'd like to throw in for the next version of
> ServiceMix:
> * documentation (we really, really need that)
> * better support for Blueprint: we should support blueprint for everyhting
> we have, it would include Camel, ActiveMQ and ServiceMix components
> * better scalability:
> - we need to put back the asynchronous support for Camel (in camel JBI
> and NMR transports)
> - we need to improve CXF-BC to better support streaming (the
> useJbiWrapper on the consumer could still use streaming as done in
> servicemix-http soap endpoints)
> - we need to find a real nice http client to have non blocking threads
> and where we would be able to write to the stream instead of having to give
> it an input stream
>
> I'll try to improve the doc on karaf and move it into a docbook format for
> easier integration into the servicemix doc, i think that would help too.
> I'll work on the Blueprint stuff, as I've already started.
> Claus is putting back async support in Camel, so once it's done, we should
> be able to put back the change we reverted for upgrading to 2.0 in both JBI
> and NMR compoennts for camel.
> CXF-BC I think still needs some work from a scalability / performance point
> of view. Having an HTTP client that could really be leveraged would be
> awesome, but I don't know any that would really work yet. I think
> httpcomponents may be a good candidate as it should be able to be used in
> asynchronous and streaming mode.
>
> Also, I think we should try to do some perfs / load testing and have the
> defaults config behave nicely, even if we need to break some compatibility
> at some point. For example, having the cxf-bc consumer being asynchronous
> by default would make sense to me.
>
> --
> Cheers,
> Guillaume Nodet
> ------------------------
> Blog: http://gnodet.blogspot.com/
> ------------------------
> Open Source SOA
> http://fusesource.com
>
--
http://lhein.blogspot.com
Re: [DISCUSS] ServiceMix RoadMap
Posted by Gert Vanthienen <ge...@gmail.com>.
L.S.,
That sounds like a good combination of goals to set for the next
release: useability (docs, blueprint, ...) and scalability.
Perhaps we could add support for Spring 3.0 to the list -- I think
Camel is already good to go there and if we move to blueprint anyway,
the effort for adding that shouldn't be too big anyway, I guess...
Regards,
Gert Vanthienen
------------------------
Open Source SOA: http://fusesource.com
Blog: http://gertvanthienen.blogspot.com/
On 17 May 2010 12:39, Guillaume Nodet <gn...@gmail.com> wrote:
> Here are a few thoughts I'd like to throw in for the next version of
> ServiceMix:
> * documentation (we really, really need that)
> * better support for Blueprint: we should support blueprint for everyhting
> we have, it would include Camel, ActiveMQ and ServiceMix components
> * better scalability:
> - we need to put back the asynchronous support for Camel (in camel JBI
> and NMR transports)
> - we need to improve CXF-BC to better support streaming (the
> useJbiWrapper on the consumer could still use streaming as done in
> servicemix-http soap endpoints)
> - we need to find a real nice http client to have non blocking threads
> and where we would be able to write to the stream instead of having to give
> it an input stream
>
> I'll try to improve the doc on karaf and move it into a docbook format for
> easier integration into the servicemix doc, i think that would help too.
> I'll work on the Blueprint stuff, as I've already started.
> Claus is putting back async support in Camel, so once it's done, we should
> be able to put back the change we reverted for upgrading to 2.0 in both JBI
> and NMR compoennts for camel.
> CXF-BC I think still needs some work from a scalability / performance point
> of view. Having an HTTP client that could really be leveraged would be
> awesome, but I don't know any that would really work yet. I think
> httpcomponents may be a good candidate as it should be able to be used in
> asynchronous and streaming mode.
>
> Also, I think we should try to do some perfs / load testing and have the
> defaults config behave nicely, even if we need to break some compatibility
> at some point. For example, having the cxf-bc consumer being asynchronous
> by default would make sense to me.
>
> --
> Cheers,
> Guillaume Nodet
> ------------------------
> Blog: http://gnodet.blogspot.com/
> ------------------------
> Open Source SOA
> http://fusesource.com
>
Re: [DISCUSS] ServiceMix RoadMap
Posted by Jean-Baptiste Onofré <jb...@nanthrax.net>.
Hi Guillaume,
my comment inline:
> * documentation (we really, really need that)
Totally agree !! We are working on it (Gert on the plugin, Charles and I
on material), but it takes time.
I will generate a first snapshot including several PDF outputs, it looks
really great.
On the same topic, I think that we need to promote the home2.html mock.
The current website is quite deprecated and doesn't really reflect the
current status of SMX 4.2. As far as I remember, Chris works on this.
> * better support for Blueprint: we should support blueprint for everyhting
> we have, it would include Camel, ActiveMQ and ServiceMix components
Agree, including components. Regarding components, I would like to add
three other enhancements:
- include OSGi unit tests to validate the deployment of a component into
SMX 4.2. It's already done in servicemix-exec but we should have it in
all components. Most of the work can be done in an abstract TestCase.
Maybe servicemix-utils can host this TestCase.
- add abstract (or concrete) WSDL publication in endpoint descriptor.
Currently servicemix-exec, servicemix-rmi (under development) and
servicemix-ldap (under development) publish a WSDL into endpoint
descriptor. As we want to "promote" the usage of Apache ODE as BPEL
engine, I would like to add WSDL support in most of components when it's
possible (I think on SMPP, XMPP, SNMP, etc).
- complete components which are in development (like servicemix-rmi and
servicemix-ldap) and add new (servicemix-atom, servicemix-jdbc and
especially the servicemix-camel-bridge).
> * better scalability:
> - we need to put back the asynchronous support for Camel (in camel JBI
> and NMR transports)
> - we need to improve CXF-BC to better support streaming (the
> useJbiWrapper on the consumer could still use streaming as done in
> servicemix-http soap endpoints)
> - we need to find a real nice http client to have non blocking threads
> and where we would be able to write to the stream instead of having to give
> it an input stream
>
> I'll try to improve the doc on karaf and move it into a docbook format for
> easier integration into the servicemix doc, i think that would help too.
> I'll work on the Blueprint stuff, as I've already started.
Great, using include, we can import the karaf doc to build a complete
SMX user guide.
> Claus is putting back async support in Camel, so once it's done, we should
> be able to put back the change we reverted for upgrading to 2.0 in both JBI
> and NMR compoennts for camel.
Is there a Camel Jira task about this ? I would like to create a SMX
Jira with the Camel one in dependency (only to avoid to forget it :)).
> CXF-BC I think still needs some work from a scalability / performance point
> of view. Having an HTTP client that could really be leveraged would be
> awesome, but I don't know any that would really work yet. I think
> httpcomponents may be a good candidate as it should be able to be used in
> asynchronous and streaming mode.
Dan already made a feedback and has very interesting idea about CXF usage.
Agree for httpcomponents, maybe we can choose a HTTP client and use it
in a whole way (in HTTP component, etc) to have only one HTTP client in
SMX and components.
>
> Also, I think we should try to do some perfs / load testing and have the
> defaults config behave nicely, even if we need to break some compatibility
> at some point. For example, having the cxf-bc consumer being asynchronous
> by default would make sense to me.
+1 we can open a Jira task about this to prepare a kind of benchmark
platform.
Regards
JB
Re: [DISCUSS] ServiceMix RoadMap
Posted by Freeman Fang <fr...@gmail.com>.
On 2010-5-17, at 下午6:39, Guillaume Nodet wrote:
> Here are a few thoughts I'd like to throw in for the next version of
> ServiceMix:
> * documentation (we really, really need that)
> * better support for Blueprint: we should support blueprint for
> everyhting
> we have, it would include Camel, ActiveMQ and ServiceMix components
> * better scalability:
> - we need to put back the asynchronous support for Camel (in
> camel JBI
> and NMR transports)
> - we need to improve CXF-BC to better support streaming (the
> useJbiWrapper on the consumer could still use streaming as done in
> servicemix-http soap endpoints)
> - we need to find a real nice http client to have non blocking
> threads
> and where we would be able to write to the stream instead of having
> to give
> it an input stream
>
> I'll try to improve the doc on karaf and move it into a docbook
> format for
> easier integration into the servicemix doc, i think that would help
> too.
> I'll work on the Blueprint stuff, as I've already started.
> Claus is putting back async support in Camel, so once it's done, we
> should
> be able to put back the change we reverted for upgrading to 2.0 in
> both JBI
> and NMR compoennts for camel.
> CXF-BC I think still needs some work from a scalability /
> performance point
> of view. Having an HTTP client that could really be leveraged would
> be
> awesome, but I don't know any that would really work yet. I think
> httpcomponents may be a good candidate as it should be able to be
> used in
> asynchronous and streaming mode.
>
> Also, I think we should try to do some perfs / load testing and have
> the
> defaults config behave nicely, even if we need to break some
> compatibility
> at some point. For example, having the cxf-bc consumer being
> asynchronous
> by default would make sense to me.
>
+1 for all, and I will start to refactor cxf-bc conusmer to use async
by default.
Freeman
> --
> Cheers,
> Guillaume Nodet
> ------------------------
> Blog: http://gnodet.blogspot.com/
> ------------------------
> Open Source SOA
> http://fusesource.com
--
Freeman Fang
------------------------
Open Source SOA: http://fusesource.com