CHAPTER 14 FLUID MODELS FOR PARALLEL PROCESSOR ALLOCATION
Dimitris Bertsimas, Eugene Perevalov · 2003
The problem of optimal processor allocation in parallel systems in the presence of nonlinear dynamics is considered. The flow of parallel jobs is modelled as that of a continuous fluid, and the Maximum Principle is applied to the resulting problem. The optimal fluid solution provides a lower bound on performance. Based on the optimal solution, several suboptimal (static partitioning) policies are proposed, and their performance is shown to be close to the lower bound in numerical examples.