DBD-Guardian and Privacy-Aware Near Real-Time Cybersecurity Analytics

Abstract

Big Data Cybersecurity Analytics (BDCA) is a helpful tool for cybersecurity maintenance that may support the identification of potential threats. Data preparation for traditional BDCA environments contains several steps comprising data movement, transformation, aggregation, and processing. All these steps take place before data becomes accessible to users. Executing such a workflow may take a reasonable time, which increases significantly with the growing amount and variety of available data sources for analytic operations. As the elapsed time between the actual occurrence of cybersecurity events and data availability for analytical queries grows, BDCA’s usefulness decreases. In this work, we deal with near real-time BDCA. We propose DBD-Guardian, a system that runs distributed queries over cybersecurity data sources (e.g., log files) while stored in their original location. DBD-Guardian supports querying heterogeneous unstructured and semi-structured sources by using specialized parsers. Also, as data sources are in their raw format, DBD-Guardian has a component specially designed to deal with sensitive data, providing access to anonymized data. To evaluate our proposals, we prototyped DBD-Guardian and implemented a representative scenario of a small company with several hosts and log files of different types. We also simulated several malicious operations in this scenario and assessed the DBD-Guardian ability to support intrusion identification and enforce privacy protection. We evaluated analytic operations’ response time as well. The results proved our solution efficiently supports analytical operations and threat identification and also demonstrated the solution’s adaptability to distributed and heterogeneous environments.

Keywords

• Cybersecurity
• near real-time analytics
• privacy
• cybersecurity analytics
• big data