I just finished watching the presentation by Jim Webber, Guerilla SOA.

From his presentation here is a few key points I drew from him:
1. ESB as a centralised infrastructure platform is really just another attempt by the software vendors to push out products. Businesses can not achieve SOA by just buying another product, and worse still the resultant ESB will be proprietary and cause further vendor-lock-in which you should try to avoid.

2. Jim then proposes the idea of “decentralised integration”. In other words the attempt to try to untangle the integration mess is possibly futile, as that is the way the business environment is, and that is analogous to the internet. So the key for SOA is really that messaging should form the key part of that infrastructure instead of an ESB. Messaging using standard SOAP envelope with policies on security, transaction etc exposed via WSDL.

Furthermore, having a centralised body for SOA governance is idealistic, and it really should be owned by the businesses who will ultimately use the application and the service. Jim quite like the idea of having a centralised wiki instead of a UDDI compliant repository where people would quickly share and collaborate information with. (Enterprise 2.0 anyone?) I think I am going to have change the direction of my basecamp project…

3. Following on the same vein, the concept of a central Service repository is also fictitious, and Jim emphasises strongly the idea of going to the owners of services for accurate data models. Instead of relying on applications to publish and update their service details in a central repository.

He also mentioned SSDL, SOAP Service Description Language, which I will probably do a bit more investigation on.

Okey, originally I just wanted to play around with Lucene, the open source search solution written in java, now I have decided to deviate from that, and go full steam ahead to create a web app that realises the enterprise 2.0 vision, i.e. blogging, wiki and social bookmarking, and everything should be searchable through this universal search function.

Since a colleague of mine has already done a lot of ground work on the above, and it happens to be written in RoR, I guess I will now use Ferret instead, which is a ruby port of Lucene.

Just set up the project in basecamp, cant wait to get cracking on it!

On top of that I also have been reading the UDDI standards, more specifically the feature list of UDDI v3, and after some googling around, it seems the software offerings around the UDDI are rather sparse… so I thought maybe we could do a port of UDDI in RoR, and finally have the full URI addressability that we have all love and talk about.

Will report back on how we go with this…

~pali

Thanks to work being rather light and not so busy, I finally get a chance to put some time to my pet project.

I have decided to invest a bit of time into Lucene, and open source search technology written in java. And create some sort of POC based on that. Have only just created the project on basecamp, so it’s still early days yet…

I just want to make some thing fun:-]