Can I extend an already created cluster?

Yes, extending a cluster is supported up to 10 nodes.

What happens if a non-primary platform VM becomes unavailable?

Internal services have limited resiliency to non-primary node failures. In general, due to a node failure NI loses compute power.

What kind of load balancing is supported?

Mapping of collector to platform is fixed. Once data from any collector VM reaches any platform VM, its processing is load-balanced internally among all the platform VMs.

Will creating a platform cluster increase bandwidth consumption?

The collector VMs continue to talk only to primary or platform VM. The bandwidth requirement for platform VM clustering communication is minimal. Therefore there is no significant increase in bandwidth consumption.

What is the frequency of data transmission between collector VM to platform VM?

The collector VM sends the deduplicated or compressed data continuously to the platform VM.

Does any optimization of data take place in the collector VM?

Various deduplications, compressions, reduction, or batching steps happen in the collector VM. When the connection between the platform VM and the collector VM is down, the collector VM stores data locally (depending on the disk space) and sends it whenever the connection is restored.

Is there any optimization done for network bandwidth?

Yes, various deduplication/compressions/reduction/batching steps happen on the collector VM.

Is clustering possible on the collector servers?

No. Clustering is not possible on the collector servers.

How does vCenter send traffic to the collector server?

vCenters do not send traffic to the collector server. The collector servers actually connect to their designated vCenter to fetch the information.

When deploying a cluster , how does vCenter send traffic to the various collector servers?

Actually proxies get connected to vCenter to fetch the information. Respective collector will connect to designated vCenter & fetch the information. There is no clustering available on collector.