TSM-Bench: methodology

Planned

Sensor data: 7 query types, queries while data streams in, compression.

Not run yet. Its queries and online mode fit. Its code has no license yet; we're asking the authors. This page is the plan.

What we'll run

We'll use its data and query templates once its license allows; timebench will run them.

The original

A benchmark from the University of Fribourg (VLDB 2023) on hydrology sensor data extended by a generator, up to 200,000 series. Seven query types: selection, filtering, aggregation, downsampling, upsampling, the average of two series, and correlation. An online mode inserts at a fixed rate while querying.

How we'll run it

Its datasets and query templates, once the license allows.
Its online mode as a timebench run that queries while ingest continues, reporting query latency against ingest rate.
Compression as bytes on storage per data point.