Lessons Learned: Continuous Deployment For Docker (presented at Cloud Expo West, 2017)
0 likes188 views
The document discusses lessons learned from implementing continuous deployment of Docker containers, focusing on the deployment of the vulnerability scanning service VCTR. Key challenges included issues with deployment scripts, which are costly to maintain and lack error handling. The document emphasizes the need for a systematic deployment approach that integrates with existing toolchains, recognizes updates, and provides automatic error correction.
Lessons Learned: Continuous Deployment For Docker (presented at Cloud Expo West, 2017)
- 1. First Hand Experience on the Path to Continuous Deployment for Docker Containers Lessons Learned: Continuous Deployment For Docker
- 2. Peter Nickolov CTO & DevOps Advocate - Opsani Opsani We help companies move their traditional apps to Docker faster without having to re-architect upfront. @Opsani_ Opsani.com github.com/opsani store.docker.com/publishers/opsani Intros @PeterNickolov github.com/pnickolov @pnickolov
- 3. App we were trying to deploy VCTR (pronounced Victor) • free service that scans containers+VMs and identifies vulnerabilities • 6,000+ Users & Integrated with PLESK
- 4. to speed up application deployment, drive down cost, ease management and make application delivery more flexible overall. We used containers to make this all easy...
- 5. What worked well - It was easy to package the app into containers - It was easy to test and qa - It was easy to create our images Create Image ProductionBuild Deploy
- 6. What didn’t work well Create Image ProductionBuild Deploy
- 7. The Story of Pavel
- 8. The Story of Pavel
- 9. The Story of Pavel
- 10. We built a system. So what did we do? Skopos
- 11. Lesson 1: Scripts Have Issues
- 12. • Prescriptive - not Declarative (like a pipeline) • Must change every time app architecture changes • Changes are multifaceted (deploy, upgrade, tear down) • Don’t take into account existing state of the app • Don’t include error handling (debugging) • Don’t clean up after themselves • Costly to maintain • No one knows what’s running now Scripts are not Features! Lesson 1: Scripts Have Issues
- 13. Lesson 2: Deployment is Systematic
- 14. 1. Integrates with popular tool chain elements and existing management systems 2. Recognizes code and architecture updates 3. Provides for identifying & respecting dependencies: start-order, statefulness 4. Provides for various operating environments: on-prem or various cloud platforms 5. Automatically deploys authorized updates (live) 6. Automatically corrects for common errors + returns to proper operation 7. Provides meaningful logs and error messages for integrated troubleshooting 8. Operates from UI, CLI, API or completely autonomously Requirements of a Deployment System
- 15. DEMO Now it’s time to see it work...
- 16. • Surveyor: See what your app looks like, share with others, convert to Docker-compose • Skopos: Deployment System for Containers • VCTR: Vulnerability Scanning for Containers Tools for You to Use No registration required. https://opsani.com/cloudexpo
- 17. Peter Nickolov CTO & DevOps Advocate - Opsani Opsani We help companies move their traditional apps to Docker faster without having to re-architect upfront. @Opsani_ Opsani.com github.com/opsani store.docker.com/publishers/opsani Thank you! @PeterNickolov github.com/pnickolov @pnickolov https://opsani.com/cloudexpo Come see us at booth 408 Power Panel: Thursday at 12:50 Speed, Quality, Impact - Building Success on the Three Foundations of DevOps