Distributed Processing

• The Computational model allows the algorithm to be parallelly performed by a cluster of phoebus nodes.
• A 'Job' submitted to a Phoebus cluster is managed by a 'Master' process running on the node that receives the Job.
• The Master partitions the input graph and spawns a 'Worker' for each partition on one of the nodes of the cluster.
• The Master then askes the Worker to perform a Super step on its partition and awaits notification from the Worker of Step completion.
• The Step number is incremented untill all Workers report that they have no more 'active' Vertices and no more outstanding messages to be deliverd.