Abstract
As organizations increasingly use cloud services to host their IT infrastructure, there is a need to share data among these cloud hosted services and systems. A majority of IT organizations have workloads spread across different cloud service providers, growing their multi-cloud environments. When an organization grows their multi-cloud environment, the threat vectors and vulnerabilities for their cloud systems and services grow as well. The increase in the number of attack vectors creates a challenge of how to prioritize mitigations and countermeasures to best defend a multi-cloud environment against attacks. Utilizing multiple industry standard risk analysis tools, we conducted an analysis of multi-cloud threat vectors enabling calculation and prioritization for the identified mitigations and countermeasures. The prioritizations from the analysis showed that authentication and architecture are the highest risk areas of threat vectors. Armed with this data, IT managers are able to more appropriately budget cybersecurity expenditure to implement the most impactful mitigations and countermeasures.
Introduction
In the early landscape of cloud computing, organizations typically consolidated their infrastructure within a single provider, such as Amazon Web Services (AWS). However, the modern enterprise has shifted toward "multi-cloud" architectures to avoid vendor lock-in and leverage specialized services across different platforms. While this approach offers flexibility, it introduces a significant interoperability challenge. The "glue" holding these disparate clouds together—primarily Application Programming Interfaces (APIs) and inter-process communication—significantly expands the attack surface. This research addresses a critical management dilemma: in a fragmented infrastructure, which vulnerabilities should be prioritized for remediation?
Methodology: Simulating the Multi-Cloud Stack
To analyze these emergent risks, the researchers developed a simulated environment based on a generic healthcare provider model. Healthcare was selected as an exemplar due to the high value of its data and the stringent regulatory requirements surrounding it, representing a "worst-case scenario" for security breaches.
The study utilized a classic three-tier web application architecture consisting of a Web Server, an App Server, and a Database (DB) Server. Crucially, this architecture was "sharded" across three distinct Cloud Service Providers (CSPs) to simulate real-world multi-cloud friction.
Risk Analysis Frameworks
The research employed a dual-layered analysis approach:
- Qualitative Analysis (STRIDE): The Microsoft-developed STRIDE model (Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of Privilege) was used to categorize potential attack vectors.
- Quantitative Analysis (DREAD): To quantify risk, the DREAD framework (Damage, Reproducibility, Exploitability, Affected users, and Discoverability) was applied. A "Total Risk Score" was calculated by averaging damage types (legal, reputation, productivity) and adding threat probabilities.
- Validation: These scores were cross-referenced against the Exploit Prediction Scoring System (EPSS) to ensure the internal risk rankings aligned with real-world threat data.
Key Results and Mitigation Strategies
The analysis revealed that the most dangerous vulnerabilities are not necessarily inherent to the individual clouds, but rather emerge from the inconsistency between them. For instance, inconsistent User Access Control Lists (ACLs) across providers can lead to unauthorized data access that would not occur in a unified environment.
The study concluded that Authentication and Architecture are the highest-risk areas in a multi-cloud setup. Architecture becomes a high-risk area because the complexity of routing traffic securely between providers is highly prone to human error.
As shown in the table above, the researchers mapped specific threats—such as VPN infiltration and session hijacking—to both technical countermeasures and administrative mitigations. A key finding was the necessity of combining technical frameworks like MITRE ATT&CK with administrative standards like ITIL (Information Technology Infrastructure Library) to manage change and secrets across providers.
Conclusion and Future Directions
By correlating DREAD risk scores with EPSS data, this methodology provides a reliable roadmap for IT managers to prioritize security budgets. The study highlights that while technical mitigations (like WAFs and encryption) are vital, administrative oversight is equally critical in a multi-cloud context.
Future research must address the "Holy Grail" of multi-cloud security: seamless identity management. Currently, identity remains fragmented across providers. Developing a method to manage a single, secure identity across AWS, Azure, and GCP—without introducing new man-in-the-middle vulnerabilities—remains the primary challenge for the next generation of cloud security.
Summary updated