Partitioned & Replicated configuration

Question

Hi,
I need to build a configuration with 6 dedicated cache nodes (ie. cache servers).
Each of these node obviously need to be outside the application environment (ie.
outside the weblogic/websphere box).
Components deployed on weblogic / websphere would invoke calls on this
cache cluster (containing the above 6 nodes).
Each node is expected to be part of a partition carrying a unique set fo data (ie.
each node will contain a different set of data). Furthermore, for the sake of
failover, each of these 6 partitioned nodes need to have a failover capability.
Can anyone post a sample configuration (XML) file for the above setup.
regards
Mike.

Answer 1

Just to re-iterate, the requirements listed are provided by the Coherence Partitioned Cache Service. See:
http://wiki.tangosol.com/display/COH32UG/Partitioned+Cache+Service
You would use a cache configuration file (XML file) to specify the cache. To enumerate the requirements and how they work:
I need to build a configuration with 6 dedicated
cache nodes (ie. cache servers). Use the built-in cache server module. See:
http://www.tangosol.com/downloads/javadoc/320/com/tangosol/net/DefaultCacheServer.html#main(java.lang.String[])
I suggest that the caches should be configured with <local-storage> as false and the cache servers should specify -Dtangosol.coherence.distributed.localstorage=true on the command line. See:
http://wiki.tangosol.com/display/COH32UG/distributed-scheme#distributed-scheme-localstorage
Each of these node obviously need to be outside the
application environment (ie.
outside the weblogic/websphere box). If (as mentioned above) the caches are configured with <local-storage> as false, then the same cache config file when used inside WebLogic or WebSphere will attach to those same caches, but the data will only be managed by the dedicated "cache servers".
Components deployed on weblogic / websphere would
invoke calls on this
cache cluster (containing the above 6 nodes).Yes, this is exactly how the Coherence Partitioned Cache Service works.
Each node is expected to be part of a partition
carrying a unique set fo data (ie.
each node will contain a different set of data).Yes, this is exactly how the Coherence Partitioned Cache Service works.
Furthermore, for the sake of
failover, each of these 6 partitioned nodes need to
have a failover capability.Yes, this is exactly how the Coherence Partitioned Cache Service works. By default, there is one level of backup (one server can die at a time without losing data). You can adjust this up or down by setting the <backup-count> element. See:
http://wiki.tangosol.com/display/COH32UG/distributed-scheme#distributed-scheme-backupcount
I hope this makes it very clear :)
Peace,
Cameron Purdy
Tangosol Coherence: Clustered Caching for Java