Monitoring Systems as a Service – Walking the Line between...
summary
Case study of migrating Udemy off unconstrained Datadog/Honeycomb usage: shows building an observability pipeline to rate‑limit dev logs, splitting retention tiers, and adding quotas; reveals 83% of custom metrics unused and a stray synthetic test costing $5k/month that gets turned off; keeps monitoring spend flat despite ~60% infrastructure growth by auditing bills, locking down admin access, automating anomaly monitors, and shifting to included/free metrics.
AI-written summary · about
chapters (9)
- 0:00Monitoring systems intro
- 3:06Vendor tools pros and cons
- 7:25Auditing monitoring costs
- 16:30Monitoring the monitor
- 17:37Log strategy for savings
- 25:01Custom metrics growth costs
- 28:10Training dev teams
- 30:19Measuring success value
- 31:43Conclusion challenges
related talks
38mDemocratizing Observability
27mBuilding a Scalable Monitoring System
46mObservability in the real world: Improve systems w/iterative approach
41mMonitoring Big Data Systems Done "The Simple Way"
39mThe Story of One Monitoring
42mMonitoring software sucks
52mBuilding a culture of observability
39mPanel Discussion: Application Monitoring