Cloudstrike: Chaos Engineering for Security and Resiliency in Cloud Infrastructure

Abstract

<jats:p>&lt;div&gt;Most cyber-attacks and data breaches in cloud&lt;/div&gt;&lt;div&gt;infrastructure are due to human errors and misconfiguration&lt;/div&gt;&lt;div&gt;vulnerabilities. Cloud customer-centric tools are lacking, and existing&lt;/div&gt;&lt;div&gt;security models do not efficiently tackle these security challenges.&lt;/div&gt;&lt;div&gt;Novel security mechanisms are imperative, therefore, we&lt;/div&gt;&lt;div&gt;propose Risk-driven Fault Injection (RDFI) techniques to tackle&lt;/div&gt;&lt;div&gt;these challenges. RDFI applies the principles of chaos engineering&lt;/div&gt;&lt;div&gt;to cloud security and leverages feedback loops to execute, monitor,&lt;/div&gt;&lt;div&gt;analyze and plan security fault injection campaigns, based on&lt;/div&gt;&lt;div&gt;a knowledge-base. The knowledge-base consists of fault models&lt;/div&gt;&lt;div&gt;designed from cloud security best practices and observations&lt;/div&gt;&lt;div&gt;derived during iterative fault injection campaigns. Furthermore,&lt;/div&gt;&lt;div&gt;the observations indicate security weaknesses and verify the&lt;/div&gt;&lt;div&gt;correctness of security attributes (integrity, confidentiality and&lt;/div&gt;&lt;div&gt;availability) and security controls. Ultimately this knowledge is&lt;/div&gt;&lt;div&gt;critical in guiding security hardening efforts and risk analysis.&lt;/div&gt;&lt;div&gt;We have designed and implemented the RDFI strategies including&lt;/div&gt;&lt;div&gt;various chaos algorithms as a software tool: CloudStrike. Furthermore,&lt;/div&gt;&lt;div&gt;CloudStrike has been evaluated against infrastructure&lt;/div&gt;&lt;div&gt;deployed on two major public cloud systems: Amazon Web Service&lt;/div&gt;&lt;div&gt;and Google Cloud Platform. The time performance linearly&lt;/div&gt;&lt;div&gt;increases, proportional to increasing attack rates. Similarly, CPU&lt;/div&gt;&lt;div&gt;and memory consumption rates are acceptable. Also, the analysis&lt;/div&gt;&lt;div&gt;of vulnerabilities detected via security fault injection has been&lt;/div&gt;&lt;div&gt;used to harden the security of cloud resources to demonstrate the&lt;/div&gt;&lt;div&gt;value of CloudStrike. Therefore, we opine that our approaches&lt;/div&gt;&lt;div&gt;are suitable for overcoming contemporary cloud security issues&lt;/div&gt;</jats:p>

Continuous auditing and threat detection in multi-cloud infrastructure

“…To provide a concrete illustration of the contemporary cloud security issues, we present a realistic running example based on the Capital One data breach [36,40]. We used this running example for a related article [55] and we have decided to reuse it here since the topics are interrelated. The running example depicts the cyber-attacks against Capital One's AWS infrastructure between April -July 2019.…”

“…• Security Chaos Engineering (SCE) [53,55] techniques are employed to validate that the changes do not introduce vulnerabilities to the cloud infrastructure. This could be done either in the development or production environment.…”

“…These is achieved by causing random actions including creation, deletion and modification of re-sources in the cloud test-bed. The fault injection is implemented with CloudStrike, a tool we designed based on the principles of chaos engineering [53,55]. CloudStrike employs three chaos modes: LOW,MEDIUM and HIGH, corresponding to random alterations in magnitudes of 30 %, 60 % and 90 % respectively, for these experiments, we used the MEDIUM chaos mode.…”