Blue-Green Deployment with Akamai

Achieving Zero-Downtime Deployments with Blue-Green Strategy and Smart Traffic Routing

---

Introduction

In modern DevOps practices, minimizing downtime and deployment risk is critical—especially for large-scale, high-traffic web applications. One approach that addresses these challenges is the blue-green deployment strategy. In this post, I’ll walk through how we designed and implemented a robust blue-green deployment workflow on a Kubernetes platform, leveraging cloud-native tools and intelligent traffic routing via CDN and load balancers. This solution enabled seamless feature rollouts, granular testing, and zero-downtime production releases.

---

The Challenge

Our team managed a major public-facing website where new features needed to be released frequently and safely. Traditional deployment strategies carried risks of downtime or inconsistent user experiences during rollouts. We needed a deployment workflow that would:

• Enable zero-downtime deployments
• Allow safe validation of new features in production-like environments
• Make rollbacks and roll-forwards straightforward
• Support targeted testing for select user groups

---

Blue-Green Deployment on Kubernetes

Our architecture ran on Amazon Elastic Kubernetes Service (EKS), with traffic entering the system via a cloud CDN and an Application Load Balancer (ALB). We adopted a blue-green deployment model as follows:

• Blue environment: The current production deployment serving live user traffic.
• Green environment: The new version, deployed in parallel but not receiving production traffic initially.

Both environments were represented as separate Kubernetes Deployments and Services.

---

Smart Traffic Routing: Cookies, Headers, and Query Strings

A key part of our solution was enabling granular, flexible traffic routing. Here’s how we accomplished this:

1. Routing via CDN and ALB

• All external traffic was routed through a CDN (e.g., Akamai, CloudFront) to our Application Load Balancer.
• The ALB supported header-based routing rules, which were used to direct requests to either the blue or green deployment in the EKS cluster.

2. User and Developer Testing

• To enable selective testing, we used query string parameters, custom cookies, or request headers.
• For example, appending a parameter like ?ENV_TARGET=green to the URL, or setting a specific cookie/header, signaled the CDN and ALB to forward the request to the green environment.
• This allowed QA engineers or business users to access the new version without exposing it to all users.

3. Workflow Example

• When a user or tester visits www.example.com?ENV_TARGET=green, the CDN sets a special header or cookie.
• The load balancer evaluates this metadata and routes the request to the green deployment.
• Similarly, requests without the override continue to route to the blue (current production) deployment.

4. Configuration Management

• Routing logic was managed using CDN property manager rules and ALB listener rules.
• Variables were set and matched (e.g., environment overrides) to ensure the right backend was selected, without affecting regular user traffic.

---

Rollout & Cutover

Once the new deployment was thoroughly tested and validated in the green environment, the final cutover was simple:

• ALB configuration was updated to direct all user traffic to the green deployment, making it the new production (“blue”).
• The previous blue deployment could be retained temporarily for instant rollback if needed.

---

Results & Impact

Implementing this strategy delivered multiple benefits:

• Zero-downtime deployments: Users experienced no interruption during releases.
• Safe rollbacks: Instantly reverted to the previous version if issues were detected.
• Targeted testing: Enabled business and QA teams to validate features before public rollout.
• Increased confidence: Streamlined our CI/CD pipeline and accelerated release cycles.

---

Key Takeaways

• Combining blue-green deployments with CDN-level routing offers powerful, flexible rollout options for cloud-native applications.
• Using cookies, headers, and query strings for environment targeting supports safe and granular testing in production-like environments.
• Automation of routing and deployment steps is key to rapid, reliable releases.

Recent posts