Optimising Mesos Utilization at Opentable
summary
OpenTable runs Mesos with Singularity and builds a metrics pipeline where Mesostats pulls Mesos/Singularity metrics into Kafka then into Graphite and Grafana dashboards with templates and inline help; dashboards show requested vs actual CPU/memory and a recommended right-size, and shifting instance types (m3→m4) cuts cloud spend ~20%. A central deployment manifest and an internal tool (sue) enforce versions; a global logging schema plus a timeline traces UUID request IDs across ~120 services and ~2–3k tasks (peaks to ~800–900 slaves), driving high Mesos adoption and real cost and ops wins.
AI-written summary · about
chapters (8)
- 0:00Start
- 2:37Busiest Day
- 8:28Development Lifecycle
- 14:10Resource Usage
- 16:06The Development Lifecycle
- 20:31Logging
- 22:41Timeline
- 27:43Key Takeaways When Deploying Missiles in the Environment
related talks
39mPractical Container Scheduling: Juggling Optimizations, Guarantees, and Trade-Offs at Netflix
12mOptimizing Multi-Tenant SaaS
12mMastering Observability in Distributed Systems
48mAPI Optimization Tale: Monitor, Fix and Deploy (on Friday)
23mIt Takes a Nation of Millions to Hold Us Back
26mGrowing Pain: Scaling to 10K Workflows per Week
34mOne Chart to Rule Them All: Continuous Deployment with Helm at Ticketmaster
26mBeyond the Hype: Facing Micro Frontend Challenges in Large Scale