Cost-Optimal Job Allocation Schemes for Bandwidth-Constrained Distributed Computing Systems

This paper formulates the job allocation problem in distributed systems with bandwidth-constrained nodes. The bandwidth limitations of the nodes play an important role in the design of cost-optimal job allocation schemes. In this paper, we present a pricing strategy for generalized distributed systems by formulating an incomplete information bargaining game on two variables (price and percentage of bandwidth allocated for distributed computing jobs at each node). Next, we present a cost-optimal job allocation scheme for single class jobs that involve the communication delay and hence link bandwidth. We show that our algorithms are comparable to existing job allocation algorithms in minimizing the expected system response time.