Two-Layer Scheduling Approach To Perform Improved Resource Utilization In Cloud Computing

Abstract

Pay-per-use virtualized resources are a key feature of the cloud computing paradigm. Scheduling is a critical part of resource management at the application and virtualization layers for cloud-scale heterogeneous network systems like those found in real life (QoS). Due to the huge solution space, cloud scheduling is generally an NP-hard issue, making it difficult to discover an optimal solution quickly. To maximise QoS parameters and meet multiple restrictions, jobs are assigned to logical resources (i.e. virtual machines) in application layer scheduling. As for application layer scheduling, a number of methods have been proposed in the literature, each of which is based on some basic technique like simple heuristics and most recently hybrid heuristics. The provisioning of resources is done in simultaneously and with consideration for Service Level Agreements (SLA). Recent research examines multiple SLA parameter techniques. The resource utilisation in the Cloud can be improved by taking into account different SLA parameters and allocating resources via a pre-emption mechanism for the execution of high priority tasks. Two-Layer Scheduling Method for Resource Allocation based on Resource Weight (TLSA-RA-RW) is a resource allocation model proposed in this research work for effectively handling streaming data in cloud environment. In Two Layer Scheduling, initially the tasks are scheduled to the virtual machines in the first layer, during execution if any task fails, the proposed model will not neglect it, rather it will allot to a active VM in the second layer. It can also be names as Two-Level Scheduling. Resource portioning is done first, followed by replication reduction, in the proposed model's design to reduce the load in the cloud environment. The resources considered are input, output, memory in the proposed model. The new model is compared to traditional models, and the results show that the proposed approach is superior to the current model.