Description
Hi Folks,
First of sorry if this not the right channel to raise queries. We are exploring memcached + mcrouter as unified caching solution for our org. One place where we are stuck (or unable to understand) is, say we are adding a new memcached pod as replica to the replica pool, from what I understand the replica will get populated gradually through the writes which will be result of cache misses on this newly added replica. If this is the case, an addition of replica will cause sudden increase in traffic on our cold store, there by we are exploring options/mechanism, if available out of the box with mcrouter, through which we can pre-populate key values on the replica and set it up for active reads. From a consistency angle, we can live with some level of inconsistency.
Thanks in advance