Currently, to mimic the clustered behavior of Redis, we utilize 2**14 segments. 
See [1]. This causes too much allocation during the startup, check [2].
 
We can create some "wrap around" our segments with a modulo operation to handle all 16K slots that Redis uses. This is necessary for clients to properly identify to which server to send a request.
 
 
[1]: https://redis.io/docs/reference/cluster-spec/#key-distribution-model
[2]: https://infinispan.zulipchat.com/#narrow/stream/118645-infinispan/topic/footprint/near/395264271