Consistency Management in Cloud Storage Systems

Abstract

With the emergence of cloud computing, many organizations have moved their data to the cloud in order to provide scalable, reliable and high available services. As these services mainly rely on geographically-distributed data replication to guarantee good performance and high availability, consistency comes into question. The CAP theorem discusses tradeoffs between consistency, availability, and partition tolerance, and concludes that only two of these three properties can be guaranteed simultaneously in replicated storage systems. With data growing in size and systems growing in scale, new tradeoffs have been introduced and new models are emerging for maintaining data consistency. In this chapter, we discuss the consistency issue and describe the CAP theorem as well as its limitations and impacts on big data management in large scale systems. We then briefly introduce several models of consistency in cloud storage systems. Then, we study some state-of-the-art cloud storage systems from both enterprise and academia, and discuss their contribution to maintaining data consistency. To complete our chapter, we introduce the current trend toward adaptive consistency in big data systems and introduce our dynamic adaptive consistency solution (Harmony). We conclude by discussing the open issues and challenges raised regarding consistency in the cloud.