A comparative analysis of resource allocation schemes for real-time services in high-performance computing systems

Abstract

An efficient resource allocation scheme plays a vital role in scheduling applications on high-performance computing resources in order to achieve desired level of service. The major part of the existing literature on resource allocation is covered by the real-time services having timing constraints as primary parameter. Resource allocation schemes for the real-time services have been designed with various architectures (static, dynamic, centralized, or distributed) and quality of service criteria (cost efficiency, completion time minimization, energy efficiency, and memory optimization). In this analysis, numerous resource allocation schemes for real-time services in various high-performance computing (distributed and non-distributed) domains have been studied and compared on the basis of common parameters such as application type, operational environment, optimization goal, architecture, system size, resource type, optimality, simulation tool, comparison technique, and input data. The basic aim of this study is to provide a consolidated platform to the researchers working on scheduling and allocating high-performance computing resources to the real-time services. This work comprehensively discusses, integrates, analysis, and categorizes all resource allocation schemes for real-time services into five high-performance computing classes: grid, cloud, edge, fog, and multicore computing systems. The workflow representations of the studied schemes help the readers in understanding basic working and architectures of these mechanisms in order to investigate further research gaps.