Budget‐constraint stochastic task scheduling on heterogeneous cloud systems

Abstract

Summary In the past few years, more and more business‐to‐consumer and enterprise applications run in the heterogeneous clouds. Such cloud bag‐of‐tasks applications are usually budget constrained, and their scheduling is an essential problem for cloud provider. The problem is even more complex and challenging when the accurate knowledge about task execution time is unknown in advance. Focusing on these challenges, we first build a cloud resource management architecture and stochastic task model, which divides cloud task into two execution parts. Then, we deduce bag‐of‐tasks applications' schedule length (Makespan) and total cost according to heterogeneous clouds' online feedback information of task first part execution. Thirdly, we formulate this stochastic scheduling problem as a linear programming problem. Lastly, we propose a time and cost multiobjective stochastic task scheduling genetic algorithm, in which can find Pareto optimal schedules for stochastic cloud task that meet its budget constraint. The extensive simulation experiments were carried out on a heterogeneous cloud platform with 400 virtual machines, and tasks were derived from Parallel Workloads Archive and the analysis data of real‐world cloud systems. The experimental results show that our proposed stochastic task scheduling genetic algorithm can get shorter schedule length and lower cost with task budget constraints.

A methodological framework for cloud resource provisioning and scheduling of data parallel applications under uncertainty

“…On the contrary, Fard et al (2016) address the uncertainty under the assumption of unknown processing time of a workflow activity by proposing a robust approach based on upper and lower bounds of processing times. Tang et al (2017) formulate a linear programming model to address the problem of scheduling tasks -whose characteristics are assumed stochastic or unknown in advance -on heterogeneous clouds under budget constraints. A probabilistic approach that takes into account uncertainties is considered in Della Vedova et al (2016a) to address the cloud resource provisioning and task scheduling of MapReduce applications.…”

Application-aware deadline constraint job scheduling mechanism on large-scale computational grid

“…Many effective heuristic and meta-heuristic Grid scheduling algorithms have been proposed to obtain near-optimal solutions, such as MET (Minimum Execution Time), Min-Min, Max-Min, and XSufferage [ 6 , 14 ]. The Min-Min heuristic algorithm tries to schedule job with overall minimum execution finish time.…”

“…Last, the newly mapped job is removed from unmapped Grid job set and the process repeats until all jobs are scheduled. The Min-Min is a traditional and widely used scheduling algorithm that has been adopted by many research works as a reference object or evaluation benchmark [ 6 , 14 , 15 , 18 ]. The improved genetic algorithm (MGA) starts with an initial population, which is generated by seeding the population with one individual generated by Min-Min, and the other individuals generated randomly.…”

High performance and security in cloud computing

“…The problem is even more complex and challenging when the accurate knowledge about task execution time is unknown in advance. Focusing on these challenges, Tang et al build a cloud resource management architecture and stochastic task model, which divides cloud task into two execution parts. Then they deduce BoT applications schedule length and total cost according to heterogenous clouds online feedback information.…”