This topic contains information on clustering WSO2 BAM versions 2.4.1 and 2.4.0.
The best architecture for BAM 2.4.0 and 2.4.1 in a high traffic scenario would be the fully-distributed highly-available setup. This basically consists of a separate data storage cluster, and also a dedicated compute cluster, which is the external Hadoop cluster for more scalable computation. So basically, in this setup, the number of nodes required for each cluster needs to be decided after doing a test on the expected load on your setup. You would also take the complexity of the analytics scripts you would have into consideration, adding more storage/compute nodes as required.