Keynote: Dockerizing legacy apps
summary
Aurea builds a central Docker PaaS on AWS aiming for 600+ containers on a single x1 instance to cut costs; actual work includes kernel upgrades to avoid unkillable D-state processes, batching iptables calls into a single iptables-restore to cut docker startup from minutes to <1 minute, and adding Kubernetes for fast recovery and orchestration. Targets are explicit: container recovery 30s, Docker API 1min, host 5min. Outcomes include operational tricks and automation that drive down footprint and enable multi-stack acquisitions at scale, with measured cost savings.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:24About Orion
- 4:27Stats
- 6:21Problems
- 8:20Symptoms
- 14:16Slow startup
- 17:14Troubleshooting
- 20:41Recovery
- 23:50Kubernetes
- 24:43Issues
- 25:45Persistent connections
- 27:13Scheme of knowledge
- 27:31Key takeaways
related talks
27mKeynote Address
58mKeynote: So, I have all these Docker containers, now what?
26mContainerization War Stories
14mKeynote: The Big Leap: Conquering the Legacy System and Building a Scalable Cloud...
52mLessons Learned from building out hyper scale cloud services using docker
08mKeynote: Cloud Native in the Enterprise: Production, Best Practice and Innovation
40mGoes Beyond CI to Production Scale PaaS with Docker
46mAround & After Kubernetes: The Principles and Ideas that Guide Us