SLA-Based Agile VM Management in Cloud & Datacenter
Nikita Sharma, Sudarshan Maurya · 2019
Cloud Computing is the main root technology to provide various services over the internet in the current scenario, which is basically powered via virtual machines (VMs) over commodity hardware. As soon as a task arrives in the cloud or data center that task is assigned to a VM as per the task scheduler and resource allocation policy. Whenever cloud or data center do not have sufficient spare resources over VM to assign a task, a new VM is created over host machine which have spare resource capacity in the cloud and/or data center. The VM creation should be in such a manner that will follow the Service Level Agreement (SLA) to maintain the Quality of Service (QoS) for the client. We have proposed SLA-based agile VM management algorithm to minimize the VM creation time and VM response time. We use ghost VM to increase the efficiency of cloud, allocation of new VM is from the ghost VM, and using the average creation time is reduced up to 11.98%. We propose two algorithms first for Admission control and second for Rescheduling of VM's. Our aim is to reduce the time taken in the rescheduling of the VM. Currently, we are assuming excess VM which are eligible to Garbage Collection do not put overhead over the cloud and they can be handled via background service, and with do not affect foreground services directly and they are handled via the separate management service.