Although XDCR provides various features that enhance or provide much flexibility to your business requirements, there is a price for enabling it in the cluster.

You need to ensure that you perform proper sizing for it, or else there will be performance impact on the existing cluster. XDCR is CPU intensive, and you need to ensure that the parallel stream number is configured based on the capacity of the server CPU. Ideally, you should have one additional core for each bucket configured for XDCR.

XDCR also doubles the I/O usage of the cluster disk. Hence, make sure that you verify the I/O capacity and increase it depending on the IOPS required for your particular application requirements. Besides disk I/O, you will require more RAM for bidirectional XDCR. You can configure an additional 20 percent of the RAM in addition to the capacity required for holding the working data set.

Here ends the discussion on the XDCR feature. Now, let's try to understand another process called compaction...