A resource allocation mechanism using coalition formation and the uncertainty principle of game theory

Abstract

Resource allocation involves assigning resources optimally and economically. In systems such as large server farms, where the demand for resources like the CPU, memory, hard-disk, or even software applications is far beyond the capabilities of a single machine, resource management involving a multitude of resources is necessary. A part of this resource management activity is resource allocation. We present a model-driven approach for optimizing resource allocation in server farms. Coalitions of servers are formed to address requests which demand capabilities beyond those of a single system. We formulate the formation of server coalitions as multiple instances of two-player zero-sum games between the servers and the task scheduler of the server farm. This two-player zero-sum game can be solved using the uncertainty principle of game theory.