aymanhs/nanotdb

NanoTDB is a small, lightweight database built for storing time series data, meaning measurements recorded over time such as temperature readings, sensor values, or device metrics. It is designed to run on low power, resource constrained hardware like Raspberry Pi computers and IoT gateways, with no external software dependencies required at runtime. All data is stored in ordinary files on disk under a single root directory rather than in a separate database server process. Data must arrive in time order, meaning each new reading has to be at the same time or later than the previous one for that sensor, and records are never changed after being written, only added on. This append only design is common in time series databases because sensor data naturally flows forward in time, and it keeps the storage engine simple and predictable. A notable feature is automatic rollup aggregation: as raw data comes in, the database can automatically compute summarized versions, such as hourly or daily averages, minimums, and maximums, and store them separately. This means that when you query data over a long time range, you can retrieve pre computed summaries instead of scanning thousands of individual readings, which makes queries much faster and lighter on the underlying hardware. The database is configured through simple text files and supports multiple isolated databases running inside a single engine instance, each with its own settings for retention and storage partitioning. It exposes an HTTP server for accepting incoming data and answering queries over that same connection. Pre built binaries are available for several Raspberry Pi models, including older and newer boards, and for 64-bit ARM systems, so you can download and run it directly without compiling any code yourself. Building from source is also possible for other platforms.