If you have a multinode cluster, which node you should connect to fetch data?

Is there a master node in a cluster which accepts data request and dispatch it? Or every node in the cluster is completely same?

If all nodes are same in a cluster, should client connect to random node to reduce cassandra's load?