Agile engineering teams often face service reliability issues primarily due to changes, as noted by Azure's CTO Mark Russinovich. To mitigate this, canary deployment is a strategy used, where a new software version is gradually exposed to a small user subset to monitor its impact, using tools like Kong Gateway and Spinnaker. At Razorpay, Kong Gateway is utilized to manage API traffic and ensure deployment reliability by splitting traffic between canary, baseline, and stable deployments. This setup allows for performance metrics comparison and rollback if anomalies are detected. Spinnaker, alongside Kayenta, further aids in automated canary analysis by scrutinizing error rates and latency, with the ability to terminate canary traffic if necessary. Kong's Admin API facilitates progressive deployments by allowing precise traffic control, starting at a low percentage and gradually increasing while monitoring performance. Additional capabilities of Kong Gateway include intelligent load balancing based on user authentication, consistent hashing to distribute traffic, and a Prometheus plugin for detailed metric analysis, aiding in the thorough evaluation of canary releases.