Thanks, Anup!!

In Cassandra when you read/write data, the request is sent to a coordinator node, this node can be any node from the cluster (based on your client config).

Now, this coordinator node sends a write/read request to the respective replica nodes based on the partition key.

Say, a Cassandra cluster has 9 nodes with a replication factor of 3. We are performing read/write requests with a consistency level of ONE.

In this case, any of the 9 nodes can be a coordinator node to handle your request and can accept read/write and then the coordinator nodes forward the read/write requests to 3 replicas based on the partition key. Once it gets acknowledgment from any one node, it will return success to the client, as consistency level is ONE.

In the context I meant, any replica can respond to a read/write request.

I have edited the section to make it more clear.