A federated system for mapreduce-based video transcoding to face the future massive video-selfie sharing trend

Abstract

The massive use of mobile devices and social networks is causing the birth of a new compulsive users’ behaviour. The activity photo selfie sharing is gradually turning into video selfie. These videos will be transcoded into multiple formats to support different visualization mode. We think there will be the need to have systems that can support, in a fast, efficient and scalable way, the millions of requests for video sharing and viewing. We think that a single Cloud Computing services provider cannot alone cope with this huge amount of incoming data (Big Data), so in this paper we propose a Cloud Federation-based system that exploiting the Hadoop MapReduce paradigm performs the video transcoding in multiple format and its distribution in a fastest and most efficient possible way. Experimental results highlight the major factors involved for job deployment in a federated Cloud environment and the efficiency of the proposed system and show how the Federation improves the performances of a MapReduce Job execution acting on a additional parallelization level.