1) Concept of Ring: Do all the nodes in a rack form a ring? Or all the nodes(every rack, every datacenter) form a ring? If the latter is true, then how are the replicas placed? let’s say I have 5 nodes, initial tokens are known.
N#1= 3
N#2=5
N#3 =7
N# 4= 9
N#5= 13
And so on.
Now, when we put a query, let’s say our key is 4, which falls in between node #1 and #2 of the ring, meaning it is present in node 2. Done. Now, we have a replication factor of 3. 3 copies are to be made. And I use RackAware strategy for it. My question is, Where are the replicas placed on the ring? When hashed, the key generated is 3. So even the replicas must have the same Key 3(that means, all the replicas are in node 2?). Does this mean that each node keeps two sets of limits - one for normal mode and one for replication ?
2) How are the nodes chosen in the other datacenter while using RackAware and DatacenterAware strategy? Does it chose any random node in the other data center? (or is there any concept of closest node while using two different Data centers?)
3) Which are the different hashing and encryption algorithms supported?
4) Can we have more than 2 Datacenters? ( if yes, how can we explain RackAware and DataCenterAware strategy?)
5) Can we configure synchronous/asynchronous replication differently for different classes of data? (ex: use asynchronous replication between different datacenters and synchronous replication the same datacenter to increase the latency)
