How We Used Jaeger and Prometheus to Deliver Lightning-Fast User Queries - Bryan Boreham

on YouTube ↗23m

" summary

Running Prometheus for cluster-wide metrics and Jaeger for distributed traces uncovers precise causes of slow user queries: tracing reveals an errored cache call, a serialization pattern (the long diagonal line indicating sequential work), and missing instrumentation around DB connections that had disabled DB connection pooling. Fixing pooling and adding spans cuts tail latency — median response time falls from roughly 22 ms to 4 ms — and the combination of metrics plus traces proves effective.

AI-written summary · about

" chapters (7)

1. 0:00Introduction
2. 2:41Three Most Important Things
3. 5:24What is Jaeger
4. 6:16The Mysterious User Service Latency
5. 11:04Jaeger
6. 15:29The Evidence
7. 18:09The Patterns

" related talks

• 45mJaeger: Present and FutureJonah Kowall, Pavol Loffay · KubeCon + CloudNativeCon Europe 2022
• 40mCracking Microservices Performance Issues with Distributed TracingDotan Horovits · ContainerDays 2022
• 35mJaeger IntroPavol Loffay, Yuri Shkuro · KubeCon + CloudNativeCon North America 2019
• 39mThe Hidden Gems of Distributed TracingAna-Maria Mihalceanu · Devnexus 2022
• 37mDeep Dive: Jaeger – Steve FlandersSteve Flanders · KubeCon + CloudNativeCon China 2019
• 51mNavigating Distributed TracesBill Westlin · QCon São Paulo 2018
• 09mGraduated Project Lightning Talk: Using Trace Data for Monitoring & Alerting of Appl...Albert Teoh · KubeCon + CloudNativeCon Europe 2021
• 37mOpenTelemetry & Distributed Tracing: A Jaeger v2 Deep DiveJonah Kowall · Open Source Monitoring Conference 2025