I have the need for two desperate Redis Instances.
One's job is mostly as a session store (currently deployed), the others role will be as an usage statistics depot. Assume my servers are single-threaded (so just adding more instances will not be a performance boon)
Currently, I am "wasting" a lot of computing resources for high availability purposes, I am wondering if this can be remedied.
My purposed architecture would be as follows (Assume a working three server sentinel configuration):
+----+ +----+ | Ma | | Mb | | Rb | | Ra | +----+ +----+ M = Master R = Slave +----+ | | = Physical Server +----+
If replication takes as much computational resources as processing requests as master, then this configuration will not do me much good on a single cpu server.
I am doing basic command, stringget, stringset, set adds, ect. Obviously, performance will be impacted if there is down time on a server, but that is not the usual case. Will this configuration let me increase throughput while utilizing more of my hardware?
Edit: There doesn't seem to be a lot of Redis activity on this site, can someone suggest a better site to migrate this question to?