I'm curious to know how far Infinispan has scaled in the real world! Anybody have stories of large numbers of nodes or huge caches?
Looking through the discussions the main thread I found was http://community.jboss.org/thread/165252?tstart=50 which mentions a presentation by Craig Bomba and Shane Johnson. Unfortunately the PDF link is not on the website http://www.redhat.com/summit/2011/presentations/jbossworld/ though it is there for some other presentations. I haven't been able to find a video of it either.
Anyone out there with large deployments in production?
I don't know why the presentation is still not there on the link you mentioned as it was submitted. I will present a link to it here if that helps. They did not video the session. Here is the link: http://www.megaupload.com/?d=YPCEPH6J
I did not see specific areas that you were interested in, but in order to reduce the amount of back and forth it might take to help you out I thought I'd mention some things that I did say from Q&A in and/or after the presentation. But first, let me say that I know where you are coming from when you want to know if the product is working reliably for anyone out there in a production environment. It is.
The current production deployment here is utilizing Replication mode. Where we have many, many nodes running Infinispan caches there are only 2 nodes per cache cluster. Cache sizes are in the range of 100,000 entries, but we have done performance testing without degredation into the millions of entries.
Hope that helps,
|Retrieving data ...|