NoSQL-based storage systems: influence of consistency on performance, availability and energy consumption

Big data applications have motivated the adoption of NoSQL database management systems (DBMS), which usually provide better performance and availability than relational DBMSs. Nowadays, these applications are commonly hosted in cloud storage services. In general, NoSQL DBMSs adopt eventual consistency, in which not all redundant nodes have the newest data, but, eventually, such data will be present in all nodes. Different levels of consistency can be utilized, but they may affect user experience and service level agreements. Therefore, techniques for evaluating the impact of eventual consistency are important for system design. This work proposes a method based on generalized stochastic Petri nets for evaluating cloud storage systems based on NoSQL DBMS using quorum technique. The models take into account distinct consistency levels and redundant nodes for estimating system availability, latency and the probability of accessing the newest data. An energy consumption model is also proposed for assessing the influence of consistency levels. Experimental results demonstrate the practical feasibility of our approach.