NanoTDB – Golang Append-Only Time Series DB
aymanhs72
54 points
11 comments
May 15, 2026
Related Discussions
Found 5 related stories in 900.1ms across 14,015 title embeddings via pgvector HNSW
- Show HN: Pg_deltax, Apache-licensed alternative to TimescaleDB tee-es-gee · 27 pts · May 19, 2026 · 48% similar
- DocumentDB – a MongoDB compatible open-source database amai · 13 pts · June 29, 2026 · 45% similar
- Show HN: Baerly-storage, a document DB that runs per request, no DB server baerbaerbaer · 15 pts · July 07, 2026 · 45% similar
- OpenData Timeseries: Prometheus-compatible metrics on object storage apurvamehta · 13 pts · April 16, 2026 · 45% similar
- How TimescaleDB compresses time-series data lkanwoqwp · 136 pts · June 15, 2026 · 45% similar
Discussion Highlights (4 comments)
xnx
How does this differ from a log file?
PunchyHamster
the history of every append only database: * we will make it append only, the type of data makes sense for it and it will simplify the design * whoops, devs fucked something up and added a bunch of nonsense that have to be removed, let's figure out how to make at least occasional deletes work
amluto
Wow, Home Assistant should try something along these lines. Home Assistant’s current handling of time series data is comically poor. Another decent option might be Clickhouse. Sadly, as far as I know, DuckDB has no real understanding of sorted or ordered data, so it might be challenging to avoid absurd amounts of read amplification.
aymanhs72
Sorry being late. Rollups were only pushed recently, as I thought about them for a long time until it finally clicked. This is main advantage over straight log files. Another things is that it is designed to be extremely efficient with SD cards. It uses WALS for keeping things safe, which is optional, and it writes very compressible data to the main database. My tests averaged 2-3 bytes per point, and the more data you have the more compression you get. So you can store many days of 10s data at less than few MBs.