Learn

How o11ytracesdb Works

A browser-native traces database takes hundreds of thousands of OpenTelemetry spans and packs them into a fraction of their original size — then answers structural queries across all of them in milliseconds.

Each experience below lets you see and interact with one piece of that puzzle. Click any card to explore.

OTLP span arrives ↓ ┌─────────────────────────┐ │ Dictionary Encoding │ strings → integer IDs │ Columnar Layout │ typed arrays per field ├─────────────────────────┤ │ Nested Set Encoding │ tree → left/right ints │ Bloom Filters │ O(1) trace ID lookup ├─────────────────────────┤ │ Structural Queries │ ancestor / descendant │ Waterfall Rendering │ span → pixel mapping └─────────────────────────┘ ↓ Answer in <1ms

Interactive Experiences

Three hands-on explorations that show how trace data gets stored, indexed, and queried.

📖

Beginner

Dictionary Encoding

See how repeated span strings (service names, HTTP methods, operations) collapse from raw UTF-8 to compact integer IDs — a 90%+ reduction.

Compression

🌳

Intermediate

Nested Set Encoding

Watch parent-child trace relationships become left/right integers that enable O(1) structural queries — no tree traversal needed.

Structure

🔍

Beginner

Bloom Filters

Add trace IDs to a bit array via hashing and see how bloom filters enable O(1) chunk-skip lookups with predictable false positive rates.

Indexing