aymanhs

aymanhs / nanotdb

Public

A tiny, append‑only time‑series database designed for long‑running sensor data on modest hardware.

99
2
89% credibility
Found May 17, 2026 at 105 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Go
AI Summary

NanoTDB is a tiny time-series database that runs directly on small computers like Raspberry Pi. It collects measurements from sensors, stores them efficiently on disk, and automatically creates summary statistics (like hourly averages) so queries are fast. Users interact through either a simple web interface or command-line tools. The database is designed to be crash-safe and requires no external services—it just stores data in plain files on your device.

How It Works

1
🏠 You have a project with sensors

Maybe it's a Raspberry Pi monitoring temperature around your house, or an IoT gateway collecting data from factory machines.

2
📦 You install NanoTDB on your device

You download and run a small program that creates a data folder. Everything stays on your device—no cloud, no accounts.

3
🔌 Your sensors start sending readings

Each sensor sends its measurements in simple text format: 'sensors/temp.office 21.5' with a timestamp. NanoTDB receives them automatically.

4
✨ Automatic summaries are created

While you sleep, NanoTDB quietly builds hourly and daily summaries—finding the minimum, maximum, and average temperatures so you don't have to crunch millions of numbers later.

5
You want to explore your data
🌐
Web browser

Open a URL like http://localhost:8428 and type queries directly in your browser, just like searching the web.

⌨️
Command line

Run simple commands to export data, inspect storage files, or run queries from your terminal.

6
📊 You ask: 'Show me last week's temperature trends'

NanoTDB scans your stored data and returns the results—either as raw readings or as pre-computed summaries, depending on what you need.

🎉 Your data is stored and organized

All your sensor readings are safely saved, automatically summarized, and ready for analysis. Your project just became a real data system.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 105 to 99 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is nanotdb?

NanoTDB is an embedded time-series database written in Go, built for long-running sensor data on modest hardware like Raspberry Pi boards, edge nodes, and IoT gateways. It stores everything in plain files under a single directory with zero runtime dependencies. You ingest data via line protocol, query it through a Prometheus-compatible HTTP API, and manage it offline with a CLI tool. The engine handles crash safety through a write-ahead log, compresses data with S2 compression (typically 3-4x on realistic sensor streams), and automatically maintains rollup aggregates so long-range queries stay fast.

Why is it gaining traction?

The automatic rollup feature is the real hook here. Instead of pre-computing aggregates in a separate pipeline, NanoTDB maintains min/max/sum/avg/count summaries at configurable intervals directly in the storage engine. You can chain rollups (1h to 1d to 1w) and they settle deterministically. The durability profiles let you trade off fsync frequency against throughput, which matters when you're running on a Pi with an SD card. Everything stays embedded and self-contained, no servers or services to babysit.

Who should use this?

Industrial IoT developers collecting sensor data on edge hardware who want persistence without Redis or InfluxDB overhead. Home automation enthusiasts running long-term logging on a Raspberry Pi. Anyone building embedded monitoring where a full database server is overkill. If you're tired of managing external services just to store temperature readings, this fits.

Verdict

NanoTDB is a well-engineered solution for its niche, but at 99 stars it's still early-stage. The documentation is thorough and the test coverage appears solid, but the credibility score of 0.9% reflects a small user base with limited real-world battle-testing. Worth evaluating for greenfield edge sensor projects, but hold off on production-critical infrastructure until the community grows.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.