Bosun: modern monitoring
summary
Presentation about Bosun, the open-source monitoring system from StackExchange, and its use at OneTwoTrip over 1.5 years. Covers problems with traditional monitoring (dashboard-watching and blanket anomaly detection), and explains Bosun’s approach of keeping metrics separate from alerting rules so teams can write, test (including on historical data), and deploy complex scenario-based alerts much faster.
AI-written summary · about
chapters (17)
- 0:00Start
- 7:51The Problem of Choice
- 9:21New Tools
- 10:01Quality Alerts
- 14:27Alert Lifecycle
- 15:48Testing on the Past
- 17:56Engaging Developers
- 21:21Alert States
- 25:22What Really Matters?
- 28:07Example Alert from OneTwoTrip
- 31:39The Story of One Outage
- 33:03OneTwoTrip and Death Queries
- 36:32Example Alert from Vimeo
- 38:13Relationships with Upstream
- 42:13Hadoop Is Not Scary
- 43:26Bosun & Chef
- 44:45Metrics from the Application
related talks
40mNext-generation Alerting and Fault Detection
39mThe Story of One Monitoring
41mMonitoring Like a Boss
28mMonitoring Stack was yesterday – Observability Platform at scale!
52mHolistic Monitoring
38m10 years of OSMC: Why does my monitoring still look the same?
23mObservability in OpenStack Learnings and building blocks from the SIG Monitoring at SCS
59mMonitoring A-Z