Operation Broken Bridge: Securing a Multi-Cloud Architecture

Cloud Security Operations
October 2025 8 Min Read

Educational Project: This operation was conducted as part of ENPM665: Cloud Security at the University of Maryland. The mission involved deploying a multi-cloud architecture (AWS backend + GCP frontend) and performing a comprehensive security posture assessment to identify and remediate critical vulnerabilities.

Multi-cloud environments offer flexibility and resilience, but they also introduce complex security challenges. In "Operation Broken Bridge," I deployed a full-stack application across AWS and GCP, then conducted a forensic after-action report to identify configuration errors, networking gaps, and security posture improvements.

2
Cloud Providers
3
Security Domains
1
Critical Config Error

Mission Overview: The Architecture

The objective was to establish a secure communication bridge between a GCP compute instance (frontend) and an AWS RDS database (backend). The architecture required precise networking configuration, secret management, and identity controls.

🌐 GCP Frontend

  • • Compute Engine (e2-micro VM)
  • • Python Flask Application (Port 5000)
  • • Firewall Rule: Allow Personal IP (129.2.89.37/32)
  • • Virtual Environment (venv) for dependencies

☁️ AWS Backend

  • • RDS PostgreSQL Instance
  • • CloudFormation Deployment (us-east-1)
  • • Security Group: Allow GCP VM Public IP (Port 5432)
  • • Static Credentials (Initial Setup)

After Action Report: What Went Well

Several critical phases of the deployment were executed successfully on the first attempt, demonstrating proficiency in cloud orchestration.

The Challenge: The "Broken Bridge"

Despite successful infrastructure deployment, the mission encountered a critical configuration error that halted connectivity between the frontend and backend.

⚠️ The Environment Variable Mismatch

The most significant time sink was a mismatch between the common convention (DB_PASSWORD) used in the .env file I created and the application's required variable name (DB_PASS).

This caused an authentication failure that presented as a "Connection Timed Out" error, delaying the diagnosis of the actual network issue. This highlights the importance of validating application requirements before infrastructure deployment.

Networking Fix: Bridging the Gap

Beyond the application configuration, the network link required precise security group tuning. The AWS Security Group's Inbound Rules had to be adjusted to allow traffic from the GCP VM's Public IP on port 5432 (PostgreSQL).

This specific task—the Phase IV "Bridge Engineering" fix—was the critical point in the architecture and should have been validated immediately after resource creation.

Security Posture Assessment

Following the successful deployment, I conducted a security audit to identify vulnerabilities in the multi-cloud environment. The assessment focused on three key domains: Secrets Management, Network Architecture, and Identity.

Security Domain Current Posture (Vulnerability) Proposed Enhancement
Secrets Management High Risk

Static, unencrypted password stored in a .env file on the VM's disk.
Zero Trust

Store password in AWS Secrets Manager. GCP VM Service Account federated with AWS IAM to retrieve short-lived passwords at runtime.
Network Architecture Exposed

GCP VM dashboard (Port 5000) and AWS RDS Security Group publicly accessible via whitelisted IPs.
Private

Implement Zero-Trust via Private IPs and IAP. GCP VM in private subnet (no public IP). Access via GCP Identity-Aware Proxy.
Identity (DB Access) Static

Connection uses traditional, long-lived username and password.
IAM Auth

Implement IAM Database Authentication. GCP VM assumes federated role, authenticates with short-lived IAM token, eliminating static DB_PASS.

Lessons Learned & Future Improvements

1. Source Code Pre-Review

Before creating the .env file, I would explicitly examine the application source code to confirm exact environment variable names (DB_HOST, DB_PASS, etc.) to prevent single-character configuration errors.

2. Strict Networking Focus

I would make the network link fix—the AWS Security Group inbound rules—the first thing I do after initial AWS deployment. This is the critical point in the "Broken Bridge" architecture.

3. Automate Deployment

To minimize human error, I would utilize the VM's Startup Script capability in GCP to automate the entire Phase III deployment (installing dependencies, configuring .env, running the app).

Conclusion

Operation Broken Bridge demonstrated the complexities of multi-cloud security. While the infrastructure was successfully deployed, the experience highlighted critical gaps in secrets management and network exposure. Moving from static credentials to federated IAM roles and from public IPs to Identity-Aware Proxy are essential steps for production-grade security.

AWS
GCP
CloudFormation
PostgreSQL
Security Groups
IAM Federation
Zero Trust

📚 Educational Context

This operation was completed as part of ENPM665: Cloud Security at the University of Maryland. The mission scenario was created for educational purposes to develop practical skills in multi-cloud deployment, network security, and security posture assessment.