We are running outbound campaigns in Pega Marketing. During the campaign dataflow execution run time, we are seeing very high cpu usage ~95% on all the batch data flow nodes. Can someone let us know what are all the settings that needs to be tweaked in order to reduce the CPU usage and also not impact the data flow runtimes. Below are the current settings we have on the system.
Total Batch Data flow nodes - 5 (all Vmware virtual servers)
CPU(s) on each node - 4
Thread(s) per core(each node) -1
Cores per socket (each node) - 1
RAM on each node - 40GB(heap 30GB)
Thread count specified on dataflow infra page - 7
MKTSegPartitionCount DSS value - 35
Batch size specified in App studio - 200
Note - We previously had used high performance physical servers(24 CPU, 2-thread/core and 6-core/socket) in place of the VMs. As a step to move towards cloud, we have replaced them with 5 VMs and we are ready to scale horizontally with additional VMs if needed.