No abstract

Due to latencies, the hadoop map reduce are complicated to scale to multiple clouds. Because of latencies, the hadoop map reduce are difficult to scale to multiple clouds. Because of this problem, to improve the performance at variable load, it provides over-provisioning in internal cloud. Here we propose a Bstream -cloud bursting framework. It consists of two major features. They are Stream-processing in the external cloud. Hadoop in the internal cloud. These two features are used to realize inter-cloud map reduce. Stream processing in external cloud enables parallel uploading; processing and also parallel downloading of data can minimize network latencies. It guarantees service-level objective (SLO) of meeting job deadlines.