Scaling Ruby Applications with Mesos and Docker
summary
A concrete migration story: about 50 existing servers and high Postgres cost lead to replacing static partitions with Apache Mesos plus Docker to treat the cluster as one resource. A JRuby 9k Sinatra service is Dockerized, published to Docker Hub, deployed via Marathon by POSTing a JSON app definition, health-checked by HTTP 200, and scaled from one to two instances on a demo mesos master with four CPUs and 2.8GB RAM. Logs go to ELK stack. Pitfalls: multi-CPU placement limits and nontrivial ops setup; Chronos handles scheduled jobs.
AI-written summary · about
chapters (10)
- 0:00Start
- 0:48Static Partitioning
- 1:23Scaling is a problem
- 3:30Resource Utilization
- 4:37Apache Mesos
- 5:42Mesos Leader & Followers
- 6:45How does Mesos work?
- 8:22Let's talk about Docker
- 15:03Mesos Scheduler
- 16:06So wha is Mesosphere?
related talks
40mDeveloping Ruby Applications with Docker
01h 01mShipping Ruby Apps with Docker
53mDocker as the Building Block for Datacenter-Scale Applications
40mContainerized Ruby Applications with Docker
47mContainerized Ruby Applications with Docker
46mDockerizing Ruby apps and moving off Engine Yard
39mHow we learned to love the Data Center Operating System
34mHow to scale a Ruby webservice