Optimizing Efficient and Accurate Real-Time Process Scheduling in Cloud Computing Systems

Abstract

One of the research problems that has to be solved is the scheduling of real-time jobs on the cloud. This problem requires taking into account the appropriate machines as well as the amount of time needed to complete the activities. The challenge with matching Realtime tasks to machines is that, assuming there are a certain number of active hosts, there are also a certain number of virtual machines (VMs) in each host. The greatest conceivable number of virtual machines that may be scheduled for a single job is (pq). In the event that we need to scheduler tasks, there are a total of (p+q)r possible combinations. Therefore, the challenge of scheduling jobs is an NP-hard problem. Real-time tasks have strict completion time constraints, and the only way for them to be helpful is if they are finished before the deadline; otherwise, they are useless. If it does not serve a helpful purpose, then it will not be accepted. The Earliest Deadline First (EDF) method is a well-known technique for scheduling activities that take place in real time. The Event Driven Framework (EDF) is a scheduling technique that assigns priorities in a dynamic manner with regard to the due dates of the tasks. There are three different types of real-time tasks: periodic, sporadic, and aperiodic. In order to evaluate the effectiveness of various scheduling methods, we have employed both aperiodic and periodic models. In most cases, the EDF Scheduler will organise the tasks in such a way that they are assigned to the free machine that is currently available. This is done without taking into account whether or not the tasks on that machine will reach the deadline. Before allocating a job to a machine, this work determined the amount of time required to do a task on all of the free machines that were available. I have utilised three distinct approaches in order to delegate the responsibility. First Fit, Best Fit, Worst Fit. Here When a task is considered fit for task, it indicates that it will successfully finish its execution on the specified machine before the deadline. The scheduling of aperiodic activities as well as periodic tasks can make use of these three approaches, in addition to the Basic EDF. The performance of the First Fit EDF (FFE), Best Fit EDF (BFE), and Worst Fit EDF (WFE) approaches has been examined by our team. The simulation was carried out in-house using the MATLAB simulator, and the performance metrics that were taken into account were the guarantee ratio (GT), virtual machine utilisation (VU), and throughput (TP). It has been demonstrated through the results of simulations that the FFE, BFE, and WFE algorithms work more effectively than the Basic EDF method.