Monitoring large scale Docker production environments
This document discusses the challenges of monitoring large scale Docker production environments. It notes that monitoring is critical for Docker in production according to 46% of respondents. When using microservices and Docker, environments can be 20 times larger, requiring techniques like network monitoring, machine-assisted problem resolution, and monitoring from the infrastructure to the application level. Effective monitoring also requires monitoring the orchestration layer, container dynamics, components like those from Netflix OSS, and the network. It should provide capabilities like visualizing the impact of automation, automated problem analysis, massive scalability, and acting as a platform feature through auto-injection and self-configuration.
Monitoring large scale Docker production environments
- 1. Monitoring large scale Docker production environments Alois Reitbauer, Dynatrace @aloisreitbauer
- 3. Docker Adoption comes in two flavors
- 4. Why do we talk about monitoring?
- 6. 46 % of all respondants see monitoring as criticial for Docker in production
- 7. Get the full report https://www.ruxit.com/docker
- 8. Why is monitoring so hard/different?
- 9. We are at the verge of a paradigm shift 90 % of 2013 best practices are about to become anti-patterns
- 10. Develop Big monolithic application. Small interconnected purpose-built services.
- 11. Two Pizza Box Teams 3 – 6 people teams can deliver features into production
- 12. New Rules in the Game You build it, you run it. Werner Vogels, CTO Amazon
- 13. Polyglot Development requires Polyglot Monitoring Wide and deep Built-in expert knowledge Monitoring is a platform feature https://vimeo.com/131385890 - James Turnmbull Monitoring as a s Service Monitorama 2015
- 14. Ship Deploy Big Bang Releases of single special built applications. Small continuous service delivery of standardized delivery blocks.
- 16. Microservice drive web-scale monitoring Environments are usually 20 times larger Network Monitoring on the rise Machine-Assisted Problem resolution From Infrastructure to Application https://www.youtube.com/watch?v=smEuX-Hq6RI - Monitoring Micro Services, Adrian Cockcroft Keynote O’Reilly Software Architecture Conference
- 18. Problems tend to become complex 820 Billion dependencies Network Problem Mushroom cloud effect
- 19. The mastery – Liquid Environments
- 20. Deployments are no longer static 7:00 a.m. Low Load and Service running on minimum redudancy 12:00 p.m. Scaled up service during peak load with failover of problematic node 7:00 p.m. Scaled down again to lower load and move to different geo location
- 21. Monitor your Orchestration Layer
- 23. Monitoring Netflix OSS Components
- 24. Network Monitoring is key
- 25. Monitoring as a Platform Feature Auto-Injection Self-configuration
- 26. Monitoring Requirements for Docker Visualize automation impact Automated problem analysis Monitor orchestration platforms Network monitoring core capability Massive scalability
- 28. Thank you! Alois Reitbauer, Dynatrace Ruxit @aloisreitbauer
Editor's Notes
- #17: 20x larger: Axel springer moved to docker about 20x more instances no changes otherwise network monitoring: networking issues just happen in such dynamic and complex cloud environments and are challenging to find needs new approaches machine-assisted: humans cannot deal with this complexity, its just too much – just like we use a navigation system to find quickly the shortest/fastest path to somewhere / use google to get results ranked by many factors to provide high likelyhood of usefulness. server monitoring is dead: it is about the entire application stack including infrastructure that serves end-users=business; and that needs monitoring