IoT-benchmark: methodology

Planned

Batched writes with out-of-order data and many clients; 12 query types; read/write mixes.

Not run yet. The best tool for mixed reads and writes. Its adapters talk to a server, so it waits for Timelake's server mode. This page is the plan.

What we'll run

We'll run its own tool, unchanged, through a Timelake adapter for its Java interface.

The original

The benchmark of Apache IoTDB, by Tsinghua University: devices with sensors write batches from many clients, with a set share of out-of-order points, while clients run 12 query types (point, range, value filter, aggregates, group by time, latest point). A ratio sets the mix of writes and each query type. It supports IoTDB, InfluxDB, TimescaleDB, QuestDB, TDengine, and others.

How we'll run it

Through a Timelake adapter for its Java interface, talking to Timelake's server mode (PostgreSQL protocol or Flight SQL).
The same settings for every system: batch size, clients, out-of-order share, and the read/write mix.