An opportunity cost approach for job assignment in a scalable computing cluster

Abstract

A new method is presented for job assignment to and reassignment between machines in a computing cluster. Our method is based on a theoretical framework that has been experimentally tested and shown to be useful in practice. This "opportunity cost" method converts the usage of several heterogeneous resources in a machine to a single homogeneous "cost". Assignment and reassignment is then performed based on that cost. This is in contrast to previous methods for job assignment and reassignment, which treat each resource as an independent entity with its own constraints. These previous methods were intrinsically ad hoc, as there was no clean way to balance one resource against another.

Model for Simulation of Heterogeneous High-Performance Computing Environments

“…Mello et al [9] have proposed improvements to the slowdown model by Amir et al [7]. This work includes new parameters which allow a better modelling of process slowdown.…”

“…Amir et al [7] have proposed a method for job assignment and reassignment on cluster computing. This method uses a queuing network model to represent the slowdown caused by virtual memory usage.…”

“…Several models have been proposed for the evaluation of the execution time and the process delay. They consider the CPU consumption, the performance slowdown due to the use of the virtual memory [7] and the time spent with messages transmitted through the communication network [8].…”

Políticas de escalonamento memory-intensive para aplicações distribuídas

“…A maioria dos trabalhos encontrados na literatura possuem métodos próprios para compararíndices de recursos heterogêneos, no entanto, o trabalho apresentado por [29] demonstra a unificação deíndices heterogêneos através do algoritmo Assign-u, convertendoos para um custo homogêneo e independente de recursos.…”

“…A segunda política proposta, utiliza os conceitos do algoritmo Assign-u [29], para homogeneizar osíndices heterogêneos e as duasúltimas utilizam o conceito de grupos, desenvolvido neste trabalho, utilizando informação tanto de CPU quanto de memória para realizar o escalonamento.…”