Data Storage and Management for Big Multimedia

This chapter focuses on the three groups of operations to implement the prototypical multimedia system: storage, processing and delivery. It discusses the commonly used big data techniques and systems for each group. The chapter focuses on media and metadata storage. It describes the storage issues in scalable cloud-based systems. The chapter analyses the current status of the storage hierarchy. The fundamental abstraction underlying most distributed file and data management systems is that of distributed hashing. In the generic system architecture presented in the introduction, media processing occurs in three places: metadata extraction, processing layer and dynamic data. The chapter presents a couple of case studies from Facebook as a means of demonstrating some of the concepts. The new field of multimedia analytics, which combines visual analytics with multimedia analysis, has been proposed to extract the knowledge and insight that large-scale multimedia collections encode in collaboration between man and machine.