Scheduling of Multimedia Over Cloud Architecture

Abstract

Cloud computing is emerging as a new computational paradigm shift. Hadoop map reduce has become powerful computational model for processing large data on distributed commodity hardware cluster such as Cloud. The strengths of map reduce are fault tolerance, an easy programming structure and high scalability. The problem of scheduling map reduce jobs are mostly caused by locality and synchronization overhead. Also, there is need to schedule multiple jobs in a share dcluste4r with farness constraints. This report reviews a collection of scheduling method for handling these issues in Map reduce. In order to implement fault tolerance in cloud environment, where nodes fail or slow down at any point, Hadoop has implemented a backup copy mechanism that will create back up copies on other nodes if something unusual happens. This method works well in homogeneous environment, but degrades in heterogeneous cluster. Hence, in this report, new backup strategy is designed that can improver job response time in heterogeneous cluster as well. The data taken to schedule is multimedia video. Multimedia data is considered to be highly challenging on cloud due to heterogeneity in terms of service, Qos, network and devices. Multimedia services and applications, such as storage and sharing, authoring and mash up, adaptation and delivery and rendering and retrieval, can optimally utilize cloud-computing recourses to achieve better Quality of Experience (QoE). There are several multimedia retrieval techniques present, whereas, we have used Content Base Video Retrieval (CBVR) technique in order to schedule the data.