Skip to main content

Benchmarks

Under construction

Andromeda has no published benchmark results yet. This page will hold them once the benchmark suite exists.

Andromeda is built to be fast. Requests are decoded straight into typed values with few allocations, the database schema is shaped around the way NGSI-LD reads and writes data, and every query is planned against indexes built for it. In testing so far it performs very well.

There are no official figures yet, and no comparison with any other broker. A speed claim only means something when anyone can repeat the measurement. The benchmark suite Andromeda will ship is meant to be reproducible and to cover many kinds of workload at many sizes, so that anyone can run it and check the figures.

Until then, try it on your own workload. Deploy it with the quickstart, load the data you actually have at the volume you expect, and run your typical writes, queries and subscriptions against it. The metrics page lists what Andromeda reports about itself.

If something is slow, open an issue with the request, the size of the data and what you expected. A slow case is treated as a bug, and a reproducible one is the best starting point the benchmark suite could have.