To load balance or cluster CAS would require that each CAS server instance in a cluster be able to validate tickets that others in the cluster issued.
There are two broad approaches to accomplishing this. One is to minimize the state that needs to be shared by an approach like making CAS tickets self-validating.
Another is to share the CAS state across CAS instances, by sharing it in an RDBMS or an in-memory Java cache using a technology like Java Groups.