Scheduling Mechanisms in Serverless Computing

Abstract

Currently, serverless computing is considered a critical event in the information and communication technologies domain. It provides a model with high scalability, pay-as-you-go, and a flexible approach to accessing requests using microservices. Many applications implement a microservice architecture, making them perform better than monolith architecture. Microservices are small pieces of code called functions, each of which can be used to run a series of processes. Consequently, microservices need the resources for execution. Hence, one of the critical issues is the efficient allocation of resources for microservices on the nodes, which are considered by structures called schedulers. Scheduling is the strategy of allocating tasks to resources in time. It increases the serverless domain’s performance and efficiency by maximizing resource utilization. This scheduling strategy has to consider restrictions specified by the serverless providers and the careers. Using the scheduler’s tasks and maintaining fairness between efficiency and the quality of careers service is complex. Scheduling algorithms are developed considering metrics similar to performance, priority, latency, cost, etc. Therefore, the process of resource allocation is regarded as a critical factor that can be of considerable significance to service providers. In serverless computing, service providers must ensure that their chosen scheduling strategies are satisfactory to service receivers. Although there are various scheduling techniques, it is crucial to point out that no single scheduling technique can accommodate all the requirements of various applicants. An in-depth understanding of the types of schedules and selecting the most effective scheduler for different kinds of applicant requirements is therefore crucial to provide the most efficient allocation of resources. In other words, when selected scheduling is inefficient, several problems can result both for the service provider and the service recipient. Therefore, service providers are forced to increase their costs. As a result, the cost of the applicants is increased. The consequences of this situation are that the requesters are dissatisfied with the poor quality service received and the increased cost they have and are less inclined to use a provider’s services in the future. Many scheduler strategies are available to providers; therefore, they should become familiar with a variety of them and understand their advantages, characteristics, and disadvantages. In this study, we comprehensively investigate the widely employed schedulers in serverless computing by investigating their advantages, disadvantages, and applications. The purpose of the present study is to present a comprehensive examination of various and effective scheduling techniques that can be a basis for selecting the appropriate scheduling process based on the providers’ approach.