Optimization of resource Allocation in cloud computing using Service based algorithms

Abstract

Cloud computing offers a lot of computing, storage and memory based resources to the users on the pay for use basis. Cloud Service Provider (CSP), users and the resources are entangled in the service cycle of cloud computing. All the Cloud computing services aim at profit and Quality of Service. The Profit of CSP depends on various factors. The foremost factor is the efficient usage of resources in cloud and it involves the process of scheduling and providing them to appropriate cloud server. Scheduling the CR (Cloud Request) is done based on the cost factors, to achieve best profit. The pricing model of cloud comes as the second factor. It is decided by the CSP, on their own standard terms and investment. Hence, optimizing the cloud based on resources and investment, plays an important role in achieving profit and efficiency. newlineInitial work deals with the deadlock prevention in Software as a Service (SaaS) cloud. An improved queuing model with Wait for resource algorithm is proposed as a solution of deadlock free cloud. For better resource utilization, the proposed model is designed to queue the arriving CRs in the first step and then a session is created between the CR and the servers with constraint on the number of CRs per session. Wait for resource algorithm pre-determines the resource needs of CR before allocation, due to this the deadlock is avoided. The proposed model gives 33 sec of average make span, 60% average resource utilization and a throughput factor of 1.3. It also gives least average waiting time and outperforms the load balancing algorithm, which produced varying response time for random applications running on cloud, without deadlock. newline