Abstract
Most cyber-attacks and data breaches in cloudinfrastructure are due to human errors and misconfigurationvulnerabilities. Cloud customer-centric tools are lacking, and existingsecurity models do not efficiently tackle these security challenges.Novel security mechanisms are imperative, therefore, wepropose Risk-driven Fault Injection (RDFI) techniques to tacklethese challenges. RDFI applies the principles of chaos engineeringto cloud security and leverages feedback loops to execute, monitor,analyze and plan security fault injection campaigns, based ona knowledge-base. The knowledge-base consists of fault modelsdesigned from cloud security best practices and observationsderived during iterative fault injection campaigns. Furthermore,the observations indicate security weaknesses and verify thecorrectness of security attributes (integrity, confidentiality andavailability) and security controls. Ultimately this knowledge iscritical in guiding security hardening efforts and risk analysis.We have designed and implemented the RDFI strategies includingvarious chaos algorithms as a software tool: CloudStrike. Furthermore,CloudStrike has been evaluated against infrastructuredeployed on two major public cloud systems: Amazon Web Serviceand Google Cloud Platform. The time performance linearlyincreases, proportional to increasing attack rates. Similarly, CPUand memory consumption rates are acceptable. Also, the analysisof vulnerabilities detected via security fault injection has beenused to harden the security of cloud resources to demonstrate thevalue of CloudStrike. Therefore, we opine that our approachesare suitable for overcoming contemporary cloud security issues