A Flow-based Task Scheduling Strategy for Distributed Systems

This paper investigates the problem of allocating a large number of independent, equal sized tasks on a distributed grid-like platform. We develop an efficient, autonomous, scalable, dynamic and generally applicable protocol for this purpose. The A-FAST protocol embodies the idea of pressure guiding the flow in fluid networks. It uses the number of unprocessed tasks buffered at each node in place of "pressure" to decide whether to move tasks to neighboring nodes. Simulations show that the A-FAST protocol performs well over a wide set of random networks, averaging more than 99% of the optimal performance. Such a protocol has the potential to aid the efficient deployment of large, data intensive applications on heterogeneous peer-to-peer computing platforms.Pre-2018 CSE ID: CS2003-074