Modern deployment strategies focus on reducing risk and ensuring a seamless user experience. By utilizing the Kubernetes Gateway API on Azure Kubernetes Service (AKS), you can gain fine-grained control over how traffic is distributed between different versions of your applications.
One of the most effective ways to manage this is through Traffic Splitting, a technique that allows you to route specific percentages of incoming requests to different backend services using weighted load balancing.
Understanding the Use Cases
Traffic splitting is the foundation for several modern DevOps patterns:
- A/B Testing: This strategy involves running two versions of an application simultaneously to see which one performs better. For example, you might test a new UI feature or marketing copy to determine which leads to higher user engagement.
- Canary Deployments: Named after the “canary in a coal mine,” this approach involves rolling out a new version to a tiny percentage of users (e.g., 5%). You monitor the performance; if everything looks good, you gradually increase the traffic until the new version reaches 100%. This limits the “blast radius” if the new version has a bug.
- Blue-Green Deployments: Here, you have two identical environments. One is “Blue” (the current production version) and the other is “Green” (the new version). Once the Green version is ready, you switch 100% of the traffic over. If an issue arises, you can instantly revert traffic back to the Blue environment.
The Power of Weighted Load Balancing
The Kubernetes Gateway API implements traffic splitting through a simple weight parameter within the HTTPRoute resource.
The weight is a positive integer that defines the relative proportion of traffic sent to a specific backend. It’s important to remember that these numbers are relative:
- A 50/50 split is the same as 1/1 or 500/500.
- A 20/80 split means 20% of traffic goes to the first service and 80% to the second.
To keep your configuration readable, it is generally recommended to use weights that add up to 100, representing percentages (e.g., 5 and 95 for a canary release).
Implementing Dynamic Traffic Shifts
One of the key advantages of using the Gateway API with Azure Application Gateway for Containers (AGFC) is the ability to update these weights on the fly.
By simply modifying the weight values in your HTTPRoute manifest and applying the change, the gateway updates its routing logic immediately. This happens without downtime; users are simply routed to the new target as soon as the configuration is processed. This responsiveness is critical for quickly rolling back a failed deployment or accelerating a successful canary rollout.
Conclusion
Traffic splitting is a foundational capability for any team looking to implement advanced deployment strategies on Kubernetes. Whether you are performing surgical A/B tests or managing high-stakes canary rollouts, the weight parameter in the Gateway API provides a simple, powerful, and reliable way to control your traffic flow. By offloading this logic to the infrastructure layer, you can focus on building better software while maintaining a safe and stable environment for your users.
You can find all the code sample on GitHub.
This post was AI-generated based on the transcript of the video “AKS Traffic Splitting Canary and A-B Deployments - Part 10” and reviewed by me.
Comments powered by Disqus.