I will not write about SCA or fabric3 in detail, maybe another time... I've "met" SCA for the first time on project, where we started with SOA. As there were some indications, that starting with ESBs is not as good shortcut as it looks, we reach for SCA. The whole project is openSource based and it is openSource itself, so we were looking for some open solution for us.
At first we started to work with Apache Tuscany (back then fabric3 was not so mature project). After couple of weeks with Tuscany we were so dissapointed with it (not just the community, but also the code itself - many bugs) that we created our own bindings (among other things) for the functional prototype, in hope, that next versions will be better. We hoped, we just change the underlying bindings from our implementation to tuscany and it will work. Well, it wasn't - still lots of problems during implementation...
Couple of months ago I was just randomly browsing on Mule community page, I think they have some SCA implementation and I remembered the fabric3. I checked their page, it looks like they've made some progress after some time and I checked their solution. Version 1.4 was the first version I started to work with, and I was amazed, what it can do. We was hoping to see that in Tuscany implementation for some time, I'm glad, we found it in fabric3.
We was looking for framework that provides infrastructure to build fault tolerance, reliability and high availability (as written on Glassfish shoal home page :-) ). Fabric3 offer standalone engine (among others) which distributed environment was in version 1.4 based on Glassfish Shoal clustering (from 1.5 it's based on JGroups). The project itself is very mature - it is the most complete SCA implementation. Also I don't think I ever worked with openSource project, where the response for some question / issue was so quick. (you can dream about it with JBoss or Apache - if you not pay for support :-) )
I'm putting together some real experience with it and I hope I will publish them here soon...
I will start with running standalone fabric3 runtimes on linux based systems (as our servers are CentOS based with some XEN virtualization).
At first we started to work with Apache Tuscany (back then fabric3 was not so mature project). After couple of weeks with Tuscany we were so dissapointed with it (not just the community, but also the code itself - many bugs) that we created our own bindings (among other things) for the functional prototype, in hope, that next versions will be better. We hoped, we just change the underlying bindings from our implementation to tuscany and it will work. Well, it wasn't - still lots of problems during implementation...
Couple of months ago I was just randomly browsing on Mule community page, I think they have some SCA implementation and I remembered the fabric3. I checked their page, it looks like they've made some progress after some time and I checked their solution. Version 1.4 was the first version I started to work with, and I was amazed, what it can do. We was hoping to see that in Tuscany implementation for some time, I'm glad, we found it in fabric3.
We was looking for framework that provides infrastructure to build fault tolerance, reliability and high availability (as written on Glassfish shoal home page :-) ). Fabric3 offer standalone engine (among others) which distributed environment was in version 1.4 based on Glassfish Shoal clustering (from 1.5 it's based on JGroups). The project itself is very mature - it is the most complete SCA implementation. Also I don't think I ever worked with openSource project, where the response for some question / issue was so quick. (you can dream about it with JBoss or Apache - if you not pay for support :-) )
I'm putting together some real experience with it and I hope I will publish them here soon...
I will start with running standalone fabric3 runtimes on linux based systems (as our servers are CentOS based with some XEN virtualization).
2 comments:
Can you give us an update on your experience with Fabric3? Did it fill your expectations or was it also disappointing at the end?
I would like to be able to contact you directly. If possible can you please contact me at juan.acosta at gmail please.
Hi Juan,
we freezed the projet based on fabric3 but just because we are focusing on different one, we will return back to it I hope in couple of months. We was able to create prototype based on fabric3 -> which was never released to production. I wouldn't say it was disappointing to work with it. It's an enterprise middleware, which is handled by metaform systems and although they don't have big community, they always responded to various requests very quickly (in comparison to e.g. jboss openSource projects). So as an opensource project it was nice experience to work with it...
Post a Comment