Securing Speed: Safeguarding CI/CD Pipelines for Robust Software Delivery

  • Some companies do “chaos engineering” to test operational resilience
  • There are lots of jobs where AGILE move fast and break things is not tenable or pragmatic
  • DevSecOps model
    • Plan, develop, build, test, release, operate, observe
  • The real phases of devsecops
    • Pre-Source Control
      • lives on computers
    • In Source Control
    • Deployment
    • Operations - is it still running?
  • Principles
    • Automate everything
    • Be lazy
    • Scale all the things
    • Measure
  • Automating helps you understand the systems better

Mission 1 - Secure the Code

  • AI makes it easy to generate bad and not scalable code
  • Humans make insecure code

Mission 2 - Secure the Repo

  • People overlook the IAM of code repos
  • Secure the Github Org
    • Public Org - allows forks, public repos, OSS
    • Secure Org - No Forks, no public repos, strict tags
  • Your code as code but with code
    • Terraform
    • IAC workflows are good - transparency - devs don’t have full admin and visibility on policies, so IAC sort of documents things
    • Control with IAC is still connected with personal access tokens in GitHub - it’s annoying but we have to live with it for now
  • Intent-based security
    • Start with intention and map abstractions to your actual intentions to be sure
  • Use strict tagging in github to drive access models and implementation
  • Branch protection
    • Ensure that reviews are required - enforce minimum security standards

https://www.youtube.com/live/vd5JpQB_owg?si=V3j4Z4QM_hYRKk5-&t=1240