A $c/\mu$-Rule for Service Resource Allocation in Group-Server Queues

TL;DR

This paper develops a $c/\mu$-rule for optimal server activation in multi-group queueing systems, establishing a structure that prioritizes groups based on cost and service rate, and demonstrating its effectiveness through sensitivity analysis.

Contribution

It introduces a $c/\mu$-rule for resource allocation in heterogeneous server groups, extending the classical $c\mu$-rule to a more complex multi-group setting with state-dependent preferences.

Findings

Optimal policy has a monotone, multi-threshold structure.

The $c/\mu$-rule effectively guides server activation priorities.

Preference order can change with system state and costs.

Abstract

In this paper, we study a dynamic on/off server scheduling problem in a queueing system with multi-class servers, where servers are heterogeneous and can be classified into $K$ groups. Servers in the same group are homogeneous. A scheduling policy determines the number of working servers (servers that are turned on) in each group at every state $n$ (number of customers in the system). Our goal is to find the optimal scheduling policy to minimize the long-run average cost, which consists of an increasing convex holding cost and a linear operating cost. We use the sensitivity-based optimization theory to characterize the optimal policy. A necessary and sufficient condition of the optimal policy is derived. We also prove that the optimal policy has monotone structures and a quasi bang-bang control is optimal. We find that the optimal policy is indexed by the value of $c - \mu G(n)$, where $c$ is the operating cost rate, $\mu$ is the service rate for a server, and $G(n)$ is a computable quantity called perturbation realization factor. Specifically, the group with smaller negative $c - \mu G(n)$ is more preferred to be turned on, while the group with positive $c - \mu G(n)$ should be turned off. However, the preference ranking of each group is affected by $G(n)$ and the preference order may change with the state $n$, the arrival rate, and the cost function. Under a reasonable condition of scale economies, we further prove that the optimal policy obeys a so-called $c$/$\mu$-rule. That is, the servers with smaller $c$/$\mu$ should be turned on with higher priority and the preference order of groups remains unchanged. This rule can be viewed as a sister version of the famous $c\mu$-rule for polling queues. With the monotone property of $G(n)$, we further prove that the optimal policy has a multi-threshold structure when the $c$/$\mu$-rule is applied.