Compensating for Failures with Flexible Servers

Abstract

We consider the problem of maximizing capacity in a queueing network with flexible servers, where the classes and servers are subject to failure. We assume that the interarrival and service times are independent and identically distributed, that routing is probabilistic, and that the failure state of the system can be described by a Markov process that is independent of the other system dynamics. We find that the maximal capacity is tightly bounded by the solution of a linear programming problem and that the solution of this problem can be used to construct timed, generalized round-robin policies that approach the maximal capacity arbitrarily closely. We then give a series of structural results for our policies, including identifying when server flexibility can completely compensate for failures and when the implementation of our policies can be simplified. We conclude with a numerical example that illustrates some of the developed insights.

Maximizing the Throughput of Tandem Lines With Flexible Failure-Prone Servers and Finite Buffers

“…Wu, Down, and Lewis [12] extend these results to serial lines with external arrivals and two stations under the discounted and average cost criteria, and develop heuristics for larger systems. Finally, Andradóttir, Ayhan, and Down [3] consider the dynamic assignment of servers to maximize the long-run average throughput of queueing networks with infinite buffers and failure-prone servers and stations. Note that both Wu, Lewis, and Veatch [13] and Wu, Down, and Lewis [12] assume that only a subset of the servers are flexible and subject to failures, and both Wu, Down, and Lewis [12] and Andradóttir, Ayhan, and Down [3] focus on systems with infinite buffers.…”

Dynamic scheduling for heterogeneous Desktop Grids

“…The Linear Programming Based Affinity Scheduling policy for Desktop Grids (LPAS DG) requires solving the following allocation LP (Andradóttir et al [3]) at each machine availability/unavailability event, where the decision variables are λ and…”

Throughput maximization for two station tandem systems: a proof of the Andradóttir–Ayhan conjecture

“…Papers representative of this approach include Harrison and Lopez [10], Bell and Williams [9], and Mandelbaum and Stolyar [13]. Andradottir et al [4,5] examined more general networks and provided heuristic round-robin policies to maximize throughput in such systems.…”