I gave a talk on our experiences clustering Confluence at the Enterprise Java Australia forums last week in Melbourne and Sydney.
The basic idea is to show our solutions to a whole host of problems that occur in clustering Java / J2EE applications so that others might learn from them. It's also known as the "stuff we really wanted to know" 6 months ago.
(Note: the presentation had to be 30 minutes including questions, so everything is very compressed right now and I ran through it at warp-speed)
I promised the attendees I'd put the slides online, so here they are (as a PDF).
I'm giving an expanded, 'full featured' version of the same presentation at Javapolis in December, but any feedback in the short term is most welcome!
I have one question:
Does any sensible person really think VB is worth drinking?
( I also have some thoughts about clustering, but I'll try and find time to put them on my blog )
Interesting deck. I wanted to point out that, although it wasn't misleading to me, people might assume that your PDF is a representative analysis of all options you list. I felt it worth noting to anyone reading this presentation that, with Terracotta:
1. Clustered caches don't have to implement Serializable interface of any kind
2. I heard from a Terracotta user that Lucene Indexes were clustered in 15 minutes
3. Clustered events work fine; as long as you have distributed wait/notify or distributed method invocation
Terracotta... they're the guys who have been spamming me with press releases for the last year or so, aren't they?
Any chance of you sharing the code for the Quartz JobStore backed by clustered cache?