Um estudo sobre interrupção e reiniciabilidade de processos em clusters de servidores Web

Abstract

With the expansion of services offered through the Internet and the popularization of the Web technology in the modern world, the toughest problems Web server administrators face are how to dimension infrastructure and how to manage the high workload during intense processing moments. As a result, we have websites with extremely high response times, due to bad resource management. This paper presents a mechanism of high availability for services and process management, for the optimization of server capacity, through a technique called process interruption and restart. The interruption and restart technique was employed to enable services to be available even through high workload moments. The restarts have three important advantages. First, it either makes a process go back to its initial state or become immune to problems. Second, the restarts have the property of increasing system reliability. The suspension and restart technique is utilized to control resource consumption under heavy processing workload. Another advantage is that restarting is a simple mechanism, easy to implement and retrieve. This has created the possibility of creating a new layer that allows better process management under severe workloads. It provides better machine performance, lower response time, and greater service availability, in addition to meeting SLA needs – QoS contracts.