> I am
> wondering what would be more latency sensitive or even bandwidth  intensive,
> Jboss Cache or database replication.

I have no idea, but it would be valuable to the community to know.  I
hope you will share the results if you take the time to measure.

> If I am doing ticket storage at the
> cluster level across datacenters, then database level replication is only
> for persistence right ?

If you choose the JpaTicketRegistry component, ticket storage _is_
persistence.  The only real issue you need to consider to make CAS
work in a cluster is ticket storage.  You can have as many nodes in
the cluster as needed, located wherever needed, as long as each shares
a common ticket store.

> So, by your last statement its best to span the
> cluster across the datacenters over TCP ?

I think having CAS server nodes in both datacenters that comprise a
single logical CAS server cluster is the way to go.  How you replicate
between them, both at the network layer (TCP, multicast) and the
application layer, will be determined by the storage technology and
your availability/capacity needs.

You're likely going to have to prototype with something and measure
the performance, and if it doesn't measure up, try something else.
You might read up on the replication overhead of JBoss Cache and
memcached so you can make a wise choice on what to try first.

M

-- 
You are currently subscribed to [email protected] as: 
[email protected]
To unsubscribe, change settings or access archives, see 
http://www.ja-sig.org/wiki/display/JSG/cas-user

Reply via email to