Oracle Data Hub Cloud Service offers a built-in capability to maintain seed nodes in the underlying Apache Cassandra cluster.
Seed nodes are a set of designated nodes in the cluster. The seed nodes are used by other nodes as a contact point to learn the cluster topology. It is responsible for:
Ensuring that there is always a minimum number of seed nodes in the cluster.
Facilitating internode communication with new nodes joining the cluster.
Avoiding single point of failure by designating multiple seed nodes in the cluster.
Preventing small clusters from being brought up due to lack of seed nodes in the cluster.
In an Oracle Data Hub Cloud Service cluster, the first three nodes are always seed nodes. During scale-out, if the new node is a multiple of 21, then the node is designated as seed node.
For example, a cluster of 50 nodes would contain 5 seed nodes. Among the 50 nodes, node number 1, 2, 3, 21, and 42 would be seed nodes.
If the scale-in operation targets a seed node, then the next adjacent node that is not a seed node is promoted to a seed node.