Publish/Subscribe-Based P2P-Cloud of Underutilized Computing Resources for Providing Computation-as-a-Service

摘要

The recent advancement in electronic devices and communication technology has lead to explosion in the digital data. These vast data have created ample opportunities to increase revenue and reduce costs, but collecting huge amount of data alone is not sufficient. To generate benefits out of these data, it must be stored and analyzed in a cost-effective and scalable manner. The general solution commonly adopted everywhere is to depend on services of cloud service provider. However, the services are costly. These days, every institute has a large number of desktop computers, laptops and other computational devices that have sufficient storage and computational power. In most of the cases, these machines are used for basic software utilities and Internet surfing which consume little amount of storage and processing power, making the resources underutilized. We aim to aggregate the underutilized computing resources available across different organizations and provide services equivalent to Computation-as-a-Service. We have developed a prototype of the framework and compared its performance with Hadoop. The experimental results suggest that in an unpredictable environment, the performance of our model is better than Hadoop.