Determination of performance of task scheduling algorithms with cloud computing technology

Cloud computing has become a current and popular technology in recent years and has become available in every field. In fact, the fact that it is encountered in every field shows why this technology is popular. Nowadays, although many devices have an internet connection, they do not have sufficient resources. What we mean by resource here is that neither processing ability, storage space nor energy source are sufficient. This is where cloud computing comes into play to solve these problems. Big data and the high complexity calculations it requires can be accessed on devices with low resources. We can define cloud computing as an internet-based computing system that provides adaptive computing resources, storage areas, servers, different applications and services without the need for interaction with the service provider and with minimum management cost. In the study, task scheduling optimization of Cloud Computing was examined using heuristic algorithms. Systems and commercial services that provide services related to Cloud Computing and open source systems such as OpenStack have been investigated. CloudSim was used for the research work. CloudSim is a simulator containing the infrastructure and services of open source cloud computing. It was developed in Java language by CLOUDS Lab. The fact that it was developed in Java is also an advantage for developers who develop software with Java. Since Java is an object-oriented language, it provides ease of use for researchers in this sense. When examining the performance of Cloud Computing in task scheduling algorithms, three factors should be taken into consideration in the resource usage of computers. Processor-intensive, storage-intensive and network traffic-intensive usage should be well organized. Namely, when processor-intensive usage occurs, there will be competition for processor slots and other resources (memory, disk, etc.) will be wasted without being used. As a result, service quality will not be high and bottlenecks will occur. Therefore, since the use of other resources will be less, there will be energy loss due to wasting of some resources. For this reason, task scheduling algorithms are needed to use resources correctly. Task scheduling algorithms are a technique used to match clients' tasks to available and appropriate virtualized resources. In heterogeneous system like cloud computing, the problem of evaluating their performance becomes more difficult as it is a distributed and scalable environment. Therefore, there is a need for an effective task scheduling algorithm, which is considered the key to understanding the performance of the system. As a task scheduling algorithm; • Traditional algorithms such as First Come First Serve (FCFS), Shortest Job First (SJF), Longest Job First (LJF) and Round Robin(RR) • Meta-heuristic algorithms such as Particle Swarm Optimization (PSO) and Gray Wolf Algorithm (GWO) Used. These algorithms were run on CloudSim with 30 inputs and their process performance was examined. As a result of the study, it was seen that meta-heuristic algorithms are superior to classical algorithms. In order for superintuitive algorithms to perform better in future studies, it is planned to create hybrid algorithms and to produce powerful hybrid algorithms by carrying out studies in this direction.