From VMs To Containers: A DevOps Journey
summary
Comcast moves a multi–data-center notification service handling hundreds of millions of publishers per day and ~500 ms delivery from VMs into Docker on Apache Mesos with Marathon, Weave networking, Jenkins pipelines and Prometheus/Graylog for observability. The containerized platform hits a puzzling 10x perf drop that traces to the Graylog GELF driver compression bug, service discovery gets solved with Consul and registrator, and the rebuilt system supports multitenancy, rolling scale and superior steady-state performance.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 4:58Application
- 8:55Decisions
- 10:57Tools
- 12:37Wargaming
- 13:07Visibility
- 13:49Performance
- 15:28Service Discovery
- 21:10The Team
- 23:08Summary
related talks
39mHow we learned to love the Data Center Operating System
48mA journey to micro-services on Mesos
49mIn Silicon Valley, Engineers Deploy-and-Tell: Stories About Dockerizing an Application
42mUsing Mesos to Drive Devops Adoption at Scale
04mPaul Cleary Interview
36mAdventures in Running a Private Cloud
25mChanging the Development Culture at Comcast and Our Evolution from Legacy SOA to Cloud Native
01h 09mDocker in Production