Metrics for High Performance Monitoring and Dashboarding - Liam McLennan - DDD Adelaide 2025 скачать в хорошем качестве

Metrics for High Performance Monitoring and Dashboarding - Liam McLennan - DDD Adelaide 2025

This talk was recorded at DDD Adelaide 2025 Imagine a particle physicist at CERN trying to log a trillion, trillion, trillion proton collisions per second. Traditional logging can't keep up, and if it could the cost would be astronomical. Where logging fails, metrics take over. While logging captures individual events in detail, metrics excel at aggregating massive volumes of high-frequency data into actionable insights. They're not just cost-effective; they're often the only practical solution for understanding systems operating at extreme scale. Metrics are increasingly accessible as a valuable and cost-effective system observability tool. This presentation explains what metrics are, when they are useful and how to go about collecting and analyzing them. Whether you're monitoring microservices, analyzing user behavior, or tracking infrastructure performance, you'll discover how metrics can unlock visibility into systems that would otherwise remain opaque. #DDDAdelaide 00:00 - DDD Adelaide logo 00:06 - Session title slide 00:11 - TxRx logo 00:17 - Introduction 01:46 - Why logs are all you need 03:25 - Why logs don't scale 05:00 - What are metrics? 08:36 - Example: Formula 1 telemetry 12:07 - Guess the metric: Ready player one 15:55 - Aggregation 18:10 - Data model 18:55 - Log event data model 20:15 - Metric data model 20:51 - Counters 25:01 - Gauges 28:25 - Summaries (Histograms) 31:21 - Summarising earthquake intensity 37:04 - Boxplot 37:45 - Violin chart 38:30 - Heatmap 39:57 - OpenTelemetry 41:20 - The sitch 41:59 - Instrumentation 43:40 - Auto instrumentation 44:20 - Log instrumentation 44:41 - Tracing instrumentation 45:10 - Metric instrumentation