Allocation of cloud resources based on prediction and performing auto-scaling of workload

Abstract

Cloud Computing allows remote access to allocated services from anywhere in the world through the internet for end users. Interpretation and analysis of real-time data are one of the most challenging tasks for cloud analysts. The determination of the correct amount of resources required to match the world is difficult. On the other hand, the large configuration makes the resource underutilized, resulting in huge economic costs. In the current decade modeling and analyzing time series data across different fields has attracted researchers in cloud computing. To overcome huge economic costs allocation of cloud resources based on prediction and performing autoscaling of workload has been proposed. This prediction analysis can avoid losses such as service unavailability, maximum energy consumption, and customer loss. When the demand is large, more resources are requested from the cloud service provider to complete the task before the deadline. When the demand is less the idle resources are released. Based on predicted values, we can reduce the workload by performing autoscaling (horizontal & vertical) in the allocation of resources.