Storage Service Reliability and Availability Predictions of Hadoop Distributed File System

Abstract

Hadoop is a de facto standard for Big Data storage and provides a complete arrangement of components for Big Data elaboration. Hadoop Distributed File System (HDFS), the fundamental module of Hadoop, has been evolved to deliver fault-tolerant data storage services in cloud. This work proposes a precise mathematical model of HDFS and estimates its data storage service availability and reliability. In this connection, a stochastic Petri net (SPN)-based dependability modelling strategy is adopted. In addition, a structural decomposition technique has been advocated to address the state space complexity of the said model. The proposed model is useful to measure crucial quality of service parameters, namely storage service reliability and availability for emerging distributed data storage systems in the context of cloud.