AWS re:Invent 2025 - Continuous integration and continuous delivery (CI/CD) for AWS (DVT202)

Continuous Integration and Continuous Delivery (CI/CD) for AWS

The Evolution of CI/CD

• Presenter discussed his experience with software development over the past 25 years, from manual build processes to the introduction of CI/CD.
• Highlighted key milestones in the evolution of CI/CD, such as the use of branches and automated deployments.
• Noted that the way we do CI/CD today will be very different in the next 3-5 years, with AI playing a significant role in transforming the process.

AWS Approach to Software Releases

• Described how AWS releases software internally, focusing on three key pillars:

  1. Gradual releases in waves to limit blast radius
  2. Automatic rollback configured in case of issues
  3. Validation and testing at each stage

• Emphasized the importance of gradual releases and working with marketing teams to manage expectations.

• Mentioned that AWS provides a full suite of tools for CI/CD, which can be mixed and matched with other tools like Jenkins.

Challenges in CI/CD Maturity

• Discussed the various pain points organizations face as they mature their CI/CD processes, including:
  • Keeping up with the rapid evolution of tools and practices
  • Balancing standardization across multiple teams
  • Introducing security vulnerabilities at any stage of the pipeline
  • Degrading pipeline performance as the codebase and number of teams grow
  • Diagnosing and reproducing pipeline failures
  • Increasing infrastructure costs as the scale grows

Simplifying CI/CD with AWS MCP

• Introduced the AWS Model Context Protocol (MCP) server, which aims to simplify the orchestration of deployments on AWS.
• Demonstrated a scenario where a developer (Joe) uses Kira, an IDE with native MCP support, to deploy a sample application to AWS.
  • Kira scans the application and generates the necessary infrastructure as code (IaC) using AWS CDK.
  • The MCP server orchestrates the deployment process, including setting up the CI/CD pipeline, connecting to the source repository, and managing the deployment to both development and production environments.
  • The MCP server also handles troubleshooting, automatically identifying and fixing issues in the pipeline.

Continuous Configuration and Feature Flagging

• Explained the concept of continuous configuration and feature flagging, and how they can be used to improve the CI/CD process.

• Highlighted the benefits of feature flagging, including:

  • Enabling trunk-based development by separating feature releases from code deployments
  • Providing fast rollback capabilities and the ability to limit the blast radius of changes
  • Allowing for validation in production and A/B testing
  • Increasing software release frequency

• Demonstrated a use case for an operational feature flag to dynamically adjust the logging verbosity based on production monitoring and anomaly detection.

Key Takeaways

• The CI/CD process is rapidly evolving, with AI playing a significant role in simplifying and automating the orchestration of deployments.
• The AWS MCP server can help teams move faster by abstracting away the complexity of setting up and managing CI/CD pipelines.
• Continuous configuration and feature flagging are powerful techniques that can improve the speed and reliability of software releases.
• Investing in these modern CI/CD practices can help organizations free up valuable time and resources to focus on innovation.