Cerved: A Journey of Innovation
summary
Cerved explains building an on-premise automated server farm that replaces VM/manual provisioning with Mesos and Marathon orchestration. The solution deploys Docker containers from Jenkins with Chef provisioning, uses Kafka to centralize logs and an ELK stack (Elasticsearch + Kibana) for log and metric collection across ~1,350 servers and ~1 PB storage, supporting ~40M LOC and ~8.5M daily transactions, with claimed ~83% faster releases and production Mesos operation.
AI-written summary · about
chapters (16)
- 0:00Start
- 10:13What emerged?
- 13:54Restart
- 15:59Why Mesos?
- 18:35Evolution of Datacenters
- 20:18MESOS Architecture
- 22:21The New Pipeline
- 24:06CERVED PIPELINE - Deploy
- 27:43The Problem of Application Logs
- 28:58MESOS - MONITORING
- 30:18NEXT STEPS
- 31:22How release times are changing
- 33:16IT vs Business Trade-offs
- 33:48New Mesos/DevOps Island
- 34:01Legacy Farm
- 34:18And so?
related talks
42mUsing Mesos to Drive Devops Adoption at Scale
39mHow we learned to love the Data Center Operating System
24mFrom VMs To Containers: A DevOps Journey
33mDeveloping for Multi-Component Environments While Keeping Your Sanity
40mTo Build My Own Cloud with Blackjack…
43mAutomated server management with open source tools
17mOur Journey to Kubernetes at Scale
47mDeploying Like a Boss: From 2 Servers to 20,000