Overview
Explore advanced techniques for root-causing latency issues in distributed systems through this 34-minute conference talk from Strange Loop. Delve into the challenges of analyzing microservice architectures and learn how recent advances in distributed tracing technology can revolutionize system analysis. Follow the evolution from Google's "Dapper" approach to modern monitoring technologies that offer full-fidelity system observation. Participate in an interactive demo connecting high-latency outlier requests to contended resources, and discover a streamlined workflow that replaces complex tooling. Gain insights into critical-path latency in distributed systems and understand how this approach can transform performance management for modern software systems.
Syllabus
Introduction
Building systems is hard
The purpose of monitoring
Microservices
Microservices and tracing
Distributed systems
Metrics
Anomalies with Metrics
Dashboards
Number of Metrics
Where is the Problem
Our Architecture
Queues
Donuts
Chasing
Sampling
Understanding Why
Hypothesis
Independent Tracing
Preview
Back to doughnuts
The doughnut zone
Graph layout library
Restock
Cleaning the fryer
Summary
Open Tracing
Taught by
Strange Loop Conference