Avaliação de Desempenho de Políticas de Escalonamento de Processos

Souza, Jorge Antonio Moraes de

doi:10.11606/d.55.2018.tde-28022018-104936

Cited by 1 publication

(1 citation statement)

References 4 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Diversas técnicas de avaliação de desempenho podem ser utilizadas para avaliar o desempenho de um escalonamento de processos. A literatura consultada discute diversas técnicas de avaliação de desempenho aplicadas ao escalonamento de processos, como por exemplo: modelagem (utilizando simulação ou técnicas analíticas para a solução dos modelos) (Buyya & Murseh, 2002;Hlavacs & Ueberhuber, 1999;Schnor et al, 1996;Souza, 2000(b); Xiao et al, 2003;Zhang et al, 2000) e benchmarks (Feitelson & Rudolph, 1998;Chapin et al, 1999). Ambas técnicas, porém, não são adequadas ao objetivo de avaliar o algoritmo de escalonamento durante a sua execução.…”

Section: Introductionunclassified

Uma abordagem para a avaliação do escalonamento de processos em sistemas distribuídos baseada em monitoração

Souza¹

View full text Add to dashboard Cite

An efficient process scheduling is of vital importance to the performance of a computer system. However, the existence of so many scheduling goals, workload types and computer system architectures makes it very difficult for a scheduling algorithm to sustain a good performance. Therefore, it is very important for the scheduling software to be capable of evaluating the computer system performance in order to guarantee the accomplishment of its scheduling goal. This thesis presents PSMS (Process Scheduling Monitoring System), an on-line monitoring system for evaluating the process scheduling. PSMS was designed to be flexible, and it is capable of communicating with any scheduler and of evaluating computer systems with different scheduling goals. The evaluation performed by PSMS is based on the analysis of performance metrics, which are chosen based on scheduling goals. Decision rules are applied on these metrics, generating a performance value that quantifies the performance of the computer system. The scheduling software has access to this performance measurement and, in case of poor performance, it can change the scheduling algorithm or perform a process migration in order to fíx the performance problem. This thesis presents a detailed description of the proposed monitoring approach and of the prototype tool developed. It also presents an analysis of performance metrics commonly used in process scheduling evaluation and a description of some decision rules. Some case studies of the use of decision rules on metrics are also discussed, demonstrating the viability of the monitoring approach proposed.

show abstract

Section: Introductionunclassified

Uma abordagem para a avaliação do escalonamento de processos em sistemas distribuídos baseada em monitoração

Souza¹

View full text Add to dashboard Cite

show abstract

Avaliação de Desempenho de Políticas de Escalonamento de Processos

Cited by 1 publication

References 4 publications

Uma abordagem para a avaliação do escalonamento de processos em sistemas distribuídos baseada em monitoração

Uma abordagem para a avaliação do escalonamento de processos em sistemas distribuídos baseada em monitoração

Contact Info

Product

Resources

About