IEEE Access (Jan 2024)
Evaluate Canary Deployment Techniques Using Kubernetes, Istio, and Liquibase for Cloud Native Enterprise Applications to Achieve Zero Downtime for Continuous Deployments
Abstract
To cater to the changing needs of the businesses, enterprises are adopting processes that allow rapid iteration and feedback loop. Today, development teams work closely with the business leveraging agile methods to gather feedback, assess the impact of the changes and deploy changes in a short duration. By taking advantage of the microservices architecture (MSA), large monolithic code is logically broken down into microservices that can be developed, deployed and scaled independently. Applications are leveraging containerization and orchestration technologies along with microservices architecture to package, deploy and manage the code across different environments. The underlying infrastructure and agile processes need to be supported with robust methods to perform code integration and deployments without any service disruption. This paper provides a qualitative assessment of the following code deployment techniques (i) Recreating Deployments (ii) Rolling Deployment (iii) Blue-Green Deployment (iv) Canary Deployment. This assessment can guide enterprises to identify the right code deployment strategy that can be adopted based on the business use case. Next, the paper dives deeper on how the in-built capabilities of Kubernetes along with open-source tools like Istio can be leveraged successfully to implement canary deployments for service changes. The paper presents a novel technique for performing canary deployments whereby service and database changes can be promoted to production by leveraging Istio and Liquibase along with load balancer without incurring any downtime for the application. The paper provides a complete canary deployment reference architecture that can be adopted by enterprises pursuing zero downtime for continuous deployments.
Keywords
