> Firstly, the data will be inserted by the coordinate node. > > Secondly, it will find the first replica node based by the partitioner ,such > randompartitioner, > > Thirdly, it will replicate the data based by the replica factor
Replicate placement is entirely independent of which node you talk to. The one talk to, the co-ordinator node, is responsible for routing the requests appropriately. The replication strategy decides where data lives. As a client, you don't have to worry about which node you're talking to except for spreading the load out over nodes in some fashion. -- / Peter Schuller
