Problem – We want to move to active-active architecture across two regions. how do we deal situation where one write happens at one region and read can happen at the other region within 10ms gap, when there are heavy write operations expected then xdcr might hit some latency in replication which leads to stale data for reads.
hence, we are thinking, Let’s create a single cluster built across the regions like - 3 nodes in one region and 3 are in other region and ensure the replicas will be on other regions with the help of server groups. I know writes and reads will hit some latency but that trade-off we are okay than the stale reads.
is this possible at all? or any other better solution available?