When a CGU is created to enforce policies, the batch processing behavior is static. That is, not until all the clusters within the first batch are compliant OR there is a timeout are any of the clusters within the next batch processed. In large batches this can waste time waiting for all clusters to complete or timeout instead of making use of the time and moving on to the next clusters. The expectation is that if the batch is size B1, then B1 number of clusters are continuously being brought into compliance.

The varying BW between different sites ultimately will cause a large variability from when the first cluster goes compliant to when the last cluster goes compliant and that time should be utilized to process clusters from the next batch.