For having a distributed topic we need to create a cluster of managed server.
Clustering comes at its own cost of constant communication between managed servers and more importantly session replication ( which Pega doesn't need).
What will be the impact if we have more than 50 JVM's ( not in cluster) connecting to the same db, however each having a regular topic and not a distributed topic?
***Edited by Moderator: Lochan to update platform capability tags***
You should always use Weblogic clustering for a production deployment (especially with so many JVms), irrespective using distributed topic or not. Pega cluster is scoped at db level, independent of Webglogic setup.
Clustering adds a lot of overhead on the JVM's as it involves a lot of peer to peer communication for session replication ( which pega doesnt need) and also communication over network via unicast/multicast.
What benefit does Pega get by having a cluster of Weblogic JVM's? What benefit does even clustering offers we dont' use that feature.
Just to give more insights, we don't use the primary and secondary jvm id WLS Clustered server for load balancing either.
So the question is if we more than 50 JVM's will there be any impact on Pega if we use a regular topic instead of a distributed topic?
Posted: 4 years ago
Posted: 22 Aug 2019 20:52 EDT
Kevin Zheng (KevinZheng_GCS)
Director, Technical Support
I see, are you running only OOTB PRAsyncTopic? It is the mechanism for all background agent processes, I can see some potential impact as messages on the topic are not shared across JVMs. I would run a comparative study with both configuration, that is the only way you can be sure which way to go based on your own metrics.