Optimizing the makespan and reliability for workflow applications with reputation and a look-ahead genetic algorithm

Abstract

For applications in large-scale distributed systems, it is becoming increasingly important to provide reliable scheduling by evaluating the reliability of resources. However, most existing reputation models used for reliability evaluation ignore the critical influence of task runtime. In addition, most previous work uses list heuristics to optimize the makespan and reliability of workflow applications instead of Genetic Algorithms (GAs), which can give several satisfying solutions for choice. Hence, in this paper, we first propose the Reliability-Driven (RD) reputation, which is time-dependent and can be used to effectively evaluate the reliability of a resource in widely distributed systems. We then propose LookAhead Genetic Algorithm (LAGA) which utilizes the RD reputation to optimize both makespan and reliability of a workflow application. LAGA uses a novel evolution and evaluation mechanism: (i) the evolution operators evolve the task-resource mapping of a scheduling solution and (ii) the evaluation step determines the task order of solutions by using our proposed max-min strategy, which is the first two-phase strategy that can work with GAs. Our experiments show that the RD reputation improves the reliability of an application with more accurate reputations, while LAGA provides better solutions than existing list heuristics and evolves to better solutions more quickly than a traditional GA.

A survey of task scheduling in Desktop Grid

“…Планирование заданий -это сложная задача, даже для случаев с полной информацией [2,[20][21][22]; в условиях реальных вычислитель-ных проектов она становится еще сложнее. Во-первых, это связано с тем, что полный набор заданий редко сформирован заранее и допол-няется динамически; во-вторых, набор доступных вычислительных узлов изменяется со временем: узлы становятся недоступны (временно или насовсем), возвращают неверные результаты, к проекту подклю-чаются новые вычислительные узлы (что особенно актуально для добровольных вычислений).…”

“…В статьях [39] и [22] предложен и исследуется прогнозный ге-нетический алгоритм для отображения заданий на узлы с целью улучшения надежности и времени обработки.…”

A Simplified Particle Swarm Optimization for Job Scheduling in Cloud Computing

“…In their study, makespan and flowtime were considered as objectives. In [35] the authors have proposed a task scheduling optimizing method based on PSO to allocate jobs to resources in a cloud so as to minimize both job computation cost and job transferring time. The authors in [36] proposed a load balancing technique based on artificial neural network (ANN), which utilizes back propagation algorithm to distribute as equally as possible the workload among all the servers.…”

Task Scheduling Using PSO Algorithm in Cloud Computing Environments

“…Wang, X., Yeo et al, [10] has proposed the Look-Ahead Genetic Algorithm (LAGA) for large-scale distributed systems such as Grid and Cloud, based on GA algorithm. It is a computation-intensive and reliability driven reputation algorithm that considers the tasks runtime using the task failure rate (task failures each unit time) of resources in order to define the status and evaluate the reliability of resources.…”

“…The inertia weight formula that was used is represented in Eq (8) [26]: (8) Where Wmin and Wmax are the minimum and maximum inertia weight values, t is the current number of iterations, the diversity function Ft and adjustment function , both in the ith iteration, are represented in equations (9) and (10), respectively: (9) Where is the group fitness as shown in equation (11): (10) Where = /3 and are the total numbers of iterations:…”