A QoS-Based Reactive Auto Scaler for Cloud Environment

Cloud computing model seems to be the preferred choice for modern day business enterprises when it comes to deploying applications. Elastic feature offered by the clouds is what is driving this transition from traditional hosting to cloud hosting. Applications hosted on clouds exhibit varying workloads, thereby making static resource provisioning less effective. Resources allocated to applications need to be tuned continuously in line with the changing workload conditions to reduce rental cost and preserve application SLAs. This paper proposes an auto-scaling model that dynamically adjusts resource allocation in a reactive manner taking into account QoS metrics. It performs resource corrections at the virtual machine level by considering both underutilization and over-utilization scenarios. The experimental results revealed the effectiveness of the proposed scheme in reducing the number of SLA violations when compared with static approach.