As EnFuzion is only aware of it's own cluster, it has no facilities to distribute files or to balance the jobs across multiple clusters. It could be done manually by copying a run and all of it's files to the other clusters, then modifying the parameters sets (jobs) on each cluster to ensure that each site does a different set. Time consuming and without knowing the speed of each cluster this may not produce the optimal results as one cluster might finish while another cluster is only half way through.
Another way to achieve multi-site with EnFuzion would be to add the nodes of the remote clusters into the local node cluster list. This will not be secure as EnFuzion uses telnet and rsh to connect to its nodes. Even then, it may not be possible as cluster installations are normally done behind a firewall. If this were not the case, this solution would work with only one other minor problem. EnFuzion allocates a job to node only after the node finishes a job. If we are now using a slow network to connect to the remote node, then the delay between jobs is increased leaving the node idle when it could be processing another job. The Multi-Site EnFuzion client overcomes this problem by allowing EnFuzion to execute normally on each cluster. There is no delay between jobs, as the local EnFuzion will quickly allocate the node a new job.
MSE will allocate the jobs across all the clusters and will continue to reshuffle jobs around to reduce completion time. There are three stages to the job allocator described in Job Allocator #1, Job Allocator #2 and Job Allocator #3.