Attack tolerant architecture for big data file systems
Bharat B. Madan, Manoj Banik · ACM SIGMETRICS Performance Evaluation Review · 2014
Data driven decisions derived from big data have become critical in many application domains, fueling the demand for collection, transportation, storage and processing of massive volumes of data. Such applications have made data a valuable resource that needs to be provided appropriate security. High value associated with big data sets has rendered big data storage systems attractive targets for cyber attackers, whose goal is to compromise the Confidentiality, Integrity and Availability of data and information. Common defense strategy for protecting cyber assets has been to first take preventive measures, and if these fail, detecting intrusions and finally recovery. Unfortunately, attackers have developed tremendous technical sophistication to defeat most defensive mechanisms. Alternative strategy is to design architectures which are intrinsically attack tolerant. This paper describes a technique that involves eliminating single point of security failures through fragmentation, coding, dispersion and reassembly. It is shown that this technique can be successfully applied to routing, networked storage systems, and big data file systems to make them attack tolerant.