> 1)   ... Store all tickets and other dynamic
> information where possible in an mysql database cluster ( replicated over
> the WAN VPN link).

I can imagine the replication would be a bandwidth hog, while latency
might be prohibitive.  The workload on the CAS ticket registry is very
read/write intensive, so you're going to generate a lot of data to
replicate.  If you wanted to have multiple nodes in the database serve
as masters where they need to receive each updates from one another, I
would think the latency would significantly reduce throughput.  Also,
I'm not up on MySQL clustering, but the only clustering setup I'm
aware of is master-slave.  Is mulit-master replication even an option?

I wonder if another ticket storage mechanism like memcached or JBoss
Cache would be more well-suited to this case.

> 2) Load balance between the data centers using technology such as Big IP's
> GTM .

You'd still have to devise a solution to share ticket state between
nodes in a cluster that spans data centers.

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