On The Margins

Thursday November 09, 2006

[ Technology ] Congratulations to Shoal

It is great to see a project I was once associated with grow, come to an amazing fruition and move forward in a wonderful direction with lots of good work behind it. I can only wish I continued to be part of such work.

Back in 2004, I had the good fortune of leading the design work for a distributed, automatic failure detection and recovery system with another member of the application server team, Shreedhar Ganapathy. He was very enthusiastic and had tried some ideas on his own already. I was charged with this because of my interest and past experience working with group communications research technologies. I think we did a really good API design and implemented a relatively good proof of concept to begin with.

I was happy we had solved some of the underlying problems having to do with the programming model for automatic recovery. However, much remained to be done. (A key issue was that all automatic failure recovery systems had at least some unaccounted-for, secondary failures that lurked behind the recovery system, for whose handling another layer of failure recovery would be needed. There is an art involved in the way you do the design work.)

I moved on to the database technologies group at Sun, Shreedhar continued with the project, combined efforts with Sun's JXTA expert and guru, Mohamed Abdel-Aziz, and now we have an amazing project worth all the attention it deserves: Shoal. Thanks to its great leveraging of the JXTA technologies, it is much  stronger and broader in its scope and effectiveness than we would have dared to dream when we started off (although I have to admit Mohamed and Shreedhar never gave up on the dreaming part, and I'm the happier for that). The added robustness in features brought by the use of JXTA is truly amazing. I'm sure the project participants look forward to continuing the work and discovering new areas for improvement...

For more useful details on the project, see blog entries by Shreedhar, Mohamed and Bernard Traversat.

It is amazing what this group has done--and that's innovation for you!

2006-11-09 15:03:18.0 -- ; Permalink ; Trackback.
cluster communications failure group recovery