While creating XDCR configuration for one my cluster in advance settings we found “XDCR Optimistic Replication Threshold” so what is the use of it and how it impacts performance. I had gone through CB documentation but not much clear , Could you please help me to understand in detail.
XDCR has the ability to check whether or not the destination already has the same or newer revision of each document before replicating it. This introduces a little bit of extra replication latency (one metadata check before sending the data) but can reduce the network bandwidth between clusters and the CPU of the destination cluster. This is particularly useful in active-active deployments with fairly large documents.
Documents larger than the “XDCR Optimistic Replication Threshold” will be checked first, documents smaller than the threshold will be sent without checking.
Are you experiencing some behaviour you think may be impacted by raising or lowering this value?
Also, this blog may be of more help: Tuning your replication stream for most optimal performance