Architecture-centric support for security orchestration and automation

Abstract

Security Orchestration, Automation and Response (SOAR) platforms leverage integration and orchestration technologies to (i) automate manual and repetitive labor-intensive tasks, (ii) provide a single panel of control to manage various types of security tools (e.g., intrusion detection system, antivirus and firewall) and (iii) streamline complex Incident Response Process (IRP) responses. SOAR platforms increase the operational efficiency of overwhelmed security teams in a Security Operation Centre (SOC) and accelerate the SOC’s defense and response capacity against ever-growing security incidents. Security tools, IRPs and security requirements form the underlying execution environment of SOAR platforms, which are changing rapidly due to the dynamic nature of security threats. A SOAR platform is expected to adapt continuously to these dynamic changes. Flexible integration, interpretation and interoperability of security tools are essential to ease the adaptation of a SOAR platform. However, most of the effort for designing and developing existing SOAR platforms are ad-hoc in nature, which introduces several engineering challenges and research challenges. For instance, the advancement of a SOAR platform increases its architectural complexity and makes the operation of such platforms difficult for end-users. These challenges come from a lack of a comprehensive view, design space and architectural support for SOAR platforms. This thesis aims to contribute to the growing realization that it is necessary to advance SOAR platforms by designing, implementing and evaluating architecture-centric support to address several of the existing challenges. The envisioned research and development activities require the identification of current practices and challenges of SOAR platforms; hence, a Multivocal Literature Review (MLR) has been designed, conducted and reported. The MLR identifies the functional and non-functional requirements, components and practices of a security orchestration domain, along with the open issues. This thesis advances the domain of a SOAR platform by providing a layered architecture, which considers the key functional and non-functional requirements of a SOAR platform. The proposed architecture is evaluated experimentally with a Proof of Concept (PoC) system, Security Tool Unifier (STUn), using seven security tools, a set of IRPs and playbooks. The research further identifies the need for and design of (i) an Artificial Intelligence (AI) based integration framework to interpret the activities of security tools and enable interoperability automatically, (ii) a semantic-based automated integration process to integrate security tools and (iii) AI-enabled design and generation of a declarative API from user query, namely DecOr, to hide the internal complexity of a SOAR platform from end-users. The experimental evaluation of the proposed approaches demonstrates that (i) consideration of architectural design decisions supports the development of an easy to interact with, modify and update SOAR platform, (ii) an AI-based integration framework and automated integration process provides effective and efficient integration and interpretation of security tools and IRPs and (iii) DecOr increases the usability and flexibility of a SOAR platform. This thesis is a useful resource and guideline for both practitioners and researchers who are working in the security orchestration domain. It provides an insight into how an architecture-centric approach, with incorporation of AI technologies, reduces the operational complexity of SOAR platforms.