This section describes how your application can connect to a storage cluster or node.

Requests to store, retrieve, or delete objects in the cluster can initially be sent to any accessible node. The cluster decides which node is best suited to carry out the request, based on resource availability and other factors.

Primary Access Node (PAN)

The node that initially fielded the request is called the PAN.

Secondary Access Node (SAN)

If the PAN did not field the request, it sends the application a redirect request that includes the address of a node referred to as the SAN. Using this method, a storage cluster performs automatic and intrinsic load balancing.

Any node in a storage cluster can serve as a PAN. The PAN for a particular request is the first node in the cluster that receives the request from the application. After a PAN receives a request, it decides whether to service the request itself or to request the application to redirect it to one of the other nodes in the cluster.

Because the PAN assumes responsibility for having the application direct the request to another node in the cluster, your application does not need to provide any load balancing solution. Swarm implements load balancing, even when nodes are dynamically added or removed from the cluster.