An intelligent cost optimized Autoscaling framework for Hybrid cloud

Abstract

Cloud computing is a revolutionary paradigm enabling on-demand provisioning of computing resources. It has shown a disruptive impact on everyday computing task. In cloud environment, the resources are delivered to cloud consumers in the form of infrastructure, platform and software services. Resource utilization plays a major role in achieving high performance, high availability and cost sustainability while adhering to Service Level Agreements (SLA). Autoscaling is one of the key features of cloud computing that has the ability to dynamically scale up or scale down the resources based on application utilization such as Central Processing Unit (CPU), Random Access Memory (RAM) and network throughput. It reduces the manual effort and provides the services quickly while adhering to SLA. The efficiency of autoscaling relies on sufficient resources offering at an optimized cost to meet the future demand. This feature is a critical aspect with any deployment model in cloud computing. newlineA hybrid cloud deployment model allows customers to host applications in multiple heterogeneous clouds. It blends the cost-effectiveness of public clouds with the control and security of private clouds. The most common deployment model for handling peak load is hybrid cloud which performs resource scaling from a private cloud to the public cloud. Although it extends the infrastructure services over various public clouds, choosing an appropriate Cloud Service Provider (CSP) based on user requirements at an optimized cost and SLAs is more complex. Different CSPs offer a wide range of flavors or instance types at different cost with same performance and capacity for heterogeneous applications. Most of the organizations map CSP and flavor to Virtual Machines (VMs) manually based on user preferences to perform resource autoscaling. Thus an effective autoscaling is mandatory in a hybrid cloud environment which requires the knowledge of each CSP offerings, application workload type, hosting environment, life time of VM, and flavor choices. newline