Now You See Me Too: Visual Tooling for Advanced System Analysis
summary
Shows how large kernel and user traces compress into actionable visuals using FlameGraphs, FlameCharts and timelines, and demonstrates concrete tooling: LTTng/CTF for capture, perf and BCC/eBPF for live stacks, and trace compass for synchronized flame/charts and a critical-path view. Examples include block I/O histograms, heat maps that reveal bimodal outliers, and hardware traces from Intel PT that expose a 173 ns slow path with ~900 extra instructions. Hardware trace rates reach 200–300 MB/s, so source filtering is essential.
AI-written summary · about
chapters (20)
- 0:00Intro
- 1:33Agenda
- 2:05Humans and Perception
- 2:49Visualizing Data
- 4:38Systems Analysis
- 5:47Visualizing Systems Data
- 9:03Observability Layers
- 10:40Data Sources
- 12:01Observing Functions
- 15:47Some Formats
- 17:17Popular Means
- 18:29Histograms
- 19:16Heatmaps
- 23:12Old Flames
- 24:40Callgraphs, Treemaps & Sunbursts
- 26:23Timelines
- 35:05Colors
- 36:25What Next?
- 40:53Future Fun
- 43:24More References
related talks
17mExtending Flamegraphs for Multi-Dimensional Performance Analysis
50mVisualizing Performance - The Developers’ Guide to Flame Graphs
44mObservability Tools C++: Beyond GDB and printf - Tools to Understand the Behavior of Your Program
33mInto the flamegraph: From the primitives through advanced concepts
01h 00mSystems @Scale Summer Q&A Week 4
26mProfiling Mastery: Decode Your Data and Elevate Observability
31mFantastic tracepoints and where to find them
37mKernel and userspace tracing with LTTng and friends