Question-56: Why it is not recommended using the replication factor of 1?

Answer: Replication factor determines how many copies of data should be kept in the cluster, suppose you have replication factor as one and the node which is holding the copy of the data goes down or have outages and cannot be re-started any more then there is no way that you can recover that lost copy.  

Question-57: Is it advisable to keep replication factor more than the number of nodes in the cluster?

Answer: You can define the replication factor with higher than the number of nodes in the cluster. If you are planning to add more nodes in the cluster. If you don't have that plan then there is no point of defining replication factor with more than number of nodes. Because Casandra does not store the same copy of the data on the same node more than one.

Question-58: What is the use of replication strategies and which one is popular for production database?

Answer: There are two replication strategy available as below

• SimpleStrategy 
• NetworkTopologyStrategy

Between this two NetworkTopologyStrategy is recommended for production environment because it is most suitable for most of the deployment. Because it is much easier to expand your cluster to multiple datacenters with this strategy. With this strategy you can define how many replicas should be placed in a particular datacenter. Suppose you have replication factor as 3 then you can decide in which datacenter you want to have one replica and, in another datacenter, you want to keep 2 replicas. 

Network topology goes clockwise in the same datacenter, and also it tries to place replica on distinct racks for having better fault-tolerant strategy.

Question-59: In the same Cassandra cluster, each keyspace can have a different replication strategy?

Answer: Yes, it is possible to have a different replication strategy for each key space. But it should be defined while creating the key space. 

Question-60: Why Murmur3Partitioner is recommended?

Answer: Purpose of having the partitioner is to derive the token such that data can be distributed evenly across the nodes in the cluster and Murmur3Partitioner is best on that. This is where the consistent hashing would helpful as well, because same logic would be used for deriving consistent hashing.