----- Original Message -----
From: "Galder Zamarreño" <galder(a)redhat.com>
To: "infinispan -Dev List" <infinispan-dev(a)lists.jboss.org>
Sent: Friday, March 2, 2012 2:01:22 PM
Subject: Re: [infinispan-dev] cross data center replication - design
Hi Mircea,
Thanks for writing that up! Some questions:
1. In the event that an admin changes the number of owners at
runtime, that'd trigger rehashing, right? Also, such change would
need to be cluster wide, right?
yes, basically the runtime change of the cluster size would be implemented with the
rehashing code and it would apply to the entire cluster.
2. What parts are you gonna be building exactly here? Any details on
how you're planning to implement this?
here are some components that would
need enhancing: transaction (be aware of sites), state transfer (to change numOwners
dinamically, to transfer all the state from a cluster to another cluster), jgroups
(bela's working on RELAY2/Bridge and we'd need to better integrate with that),
administration and of course testing. There's also a new routing component that needs
to be developed that would run on each bridge end and would know how to direct requests
(e.g. stat transfer, prepare) to the appropriate nodes in the local cluster.