Hi Janne,
one important reason is that JBI is more or less dead. Besides
servicemix there is not much support from vendors anymore. The spec also
does not seem to evolve. There was a discussion on the dev list that
will explain more:
http://servicemix.396122.n5.nabble.com/DISCUSS-ServiceMix-future-td3212177.html
The even more important reason to use OSGi is that it is not limited to
ESBs. You can write Eclipse RCP GUIs, Server based apps using e.g. Karaf
and of course integrations using Servicemix. So it is much
easier to leverage existing libraries. It is also much more proable that
your own libraries are reused.
Another reason is that it is quite easy to write OSGi bundles in a way
so they can also be used standalone or in a web container. All examples
I did for Talend Integration factory run standalone, as a war and in
OSGi. All this works with the same source. So if parts of your
enterprise prefer to deploy in a web container you can still use the
same tools and share sources.
So while I am quite sure that the JBI packaging is dead I am not so sure
about the Normalized Message Router part of Servicemix. I did not yet
have to use it for my integrations but I can imagine it can be handy.
Best regards
Christian
Am 16.02.2011 13:17, schrieb janne postilista:
Hi,
why would I prefer OSGi over JBI (and is it a question of choosing
either)? I thought OSGi was more or less just a way of packaging a JBI
service assembly (but maybe its not...)?
I thought JBI was a good thing (standardized packaging, common
concepts in all supporting ESBs, etc)? Why would I not want to develop
JBI artifacts? Is JBI considered bad for some reasons? If I develop
"simple osgi bundles", am I not tied into servicemix tighter than if I
develop JBI sa's (then I can move them more easily to any JBI
compliant ESB)?
--
----
http://www.liquid-reality.de
