Zero* Cost Observability: A Deep Dive into the Jetpack Multiplatform Tracing Library

on YouTube ↗31m

" summary

Library claims 10x faster than legacy ATrace (1.2 ms down to ~0.12 ms on Pixel 6A), emits Perfetto-format protos and supports swapping in custom back ends, and is KMP-ready for JVM + Android. Implementation uses zero allocations in the hot path via Wire mutable types and proto3, offloads serialization with coroutine intrinsics to reuse Dispatchers IO, and provides context propagation for coroutines plus a propagation token for callbacks. Benchmarks drive changes (atomics replaced, specialized collections, object pooling) and the end-to-end system demonstrably works in traces shown.

AI-written summary · about

" chapters (11)

1. 0:00Tracing on Android
2. 0:45Prior tracing API limits
3. 5:18New tracing 2.0 library
4. 8:03Tracing API basics
5. 9:26Metadata in trace events
6. 10:08Overlapping and nested events
7. 11:12Call stack in traces
8. 11:54Async context propagation
9. 16:04Zero-cost tracing abstractions
10. 26:54Performance benchmarks
11. 29:43Resources and closing

" related talks

• 25mPerfect Match: Correlating Continuous Profiling with Distributed Tracing for Stronger ObservabilityChristos Kalkanis, Jonas Kunz · Observability Day North America 2024
• 34mClient Tracing At ScaleBen Romano · droidcon San Francisco 2022
• 47mPerformance in a Kotlin WorldSean McQuillan · Android Summit 2019
• 45mPerformance in a Kotlin worldSean McQuillan · Chicago Roboto 2020
• 34mAdventures building a Kotlin Multiplatform Benchmarking LibraryRahul Ravikumar · KotlinConf 2023
• 43mPerformance in a Kotlin WorldSean McQuillan · droidcon SF 2019
• 04mSponsored Keynote: Zero-Code Observability: Close the Coverage Gaps That Cause Outa...Eden Federman · Observability Summit North America 2026
• 19mKeynote: OpenTracing and Containers: Depth, Breadth, and the Future of Tracing - Ben SigelmanBen Sigelman · KubeCon + CloudNativeCon North America 2016