Hi,
If you use different RR´s using the same cluster ID, a RR will reject prefixes received from a route reflector due to the loop avoidance mechanism. Therefore, you have less redundancy in comparison to use a different Cluster ID per RR. The drawback of using a different Cluster ID per RR is an increase in memory requirements due to the number of prefixes received (from Clients plus from RRs).
Hope this helps,
Jose.