Resource provisioning technique for memory-intensive cloud applications

Abstract

Cloud Computing is gaining popularity in IT industry due to the increase in demand of the computing resources as a service. Cloud Computing reduces the operational and computational cost and thus large companies get benefitted in a manner as they need to pay for as much amount of resource as they need to carry out their present business. With the increase in the requirement they can pay more and get more resources for them, and thus according to their need they can scale up and scale down the resources. Cloud service users require the specific Quality of Service (QoS) in order to meet their objective and sustain their operations on the basis of the Service Level Agreements (SLAs). Resource provisioning in cloud computing deals with how the resources may be allocated to the application mix such that the SLAs of all applications are met. SLAs may vary from user to user and in order to fulfill the specific SLA it is necessary for the cloud provider to manage the resources efficiently. Resource provisioning allows the users and providers to access the specified resources according to availability of the resources in Cloud as per their request. There is a need to improve the optimization and allocation of cloud resources. Efficient algorithms for resource optimization are required to be developed for this purpose. There is also a need optimize the resources for memory intensive Cloud applications. In this thesis the existing techniques of resource provisioning on Cloud have been compared and a technique to optimize the memory intensive applications on Cloud infrastructure has been proposed. The validation of the proposed techniques is done on the Eucalyptus Testbed which helps in the scaling of the resources. The proposed technique has been validated through RUBiS application deployed on EUCALYPTUS cloud set up. The experimental results demonstrate that the proposed technique helps to scale up and scale down the cloud infrastructure as per user requirement and thus providing efficient resource provisioning.