How are requests spread across a Swarm cluster?

Any node can be contacted to fulfill a request. This node is referred to as the Primary Access Node (PAN). The PAN will analyze the request and if necessary redirect the client (or SCSP Proxy/ Content Gateway) to the Swarm node that is in the best position to service the request. The nature of this process will ensure that all requests to a Swarm cluster are distributed across the whole cluster evenly. To eliminate the possibility that a single node is used as the PAN for all requests we recommend that the client application use multiple nodes in the cluster as the PAN. One way to do this is that upon completing a request the next request can be made to the last node to which they communicated. This way as the client is redirected to nodes to service requests, that node now becomes the PAN for next request and so on. Doing this, requests will move around the entire cluster and eliminate the possibility that a single node is used as the PAN.