Search under load

Local search can feel fast with a small dataset. This experiment asks how that experience changes as more content is indexed and more queries compete for resources.

The documented focus is SQLite full-text retrieval with FTS5. The case examines dataset size and query pressure as separate variables so a slowdown can be connected to the conditions that produced it.

A useful comparison needs a fixed query set, a recorded dataset size, and consistent execution conditions. Warm and cold runs should be distinguished, and latency should be reported as a distribution rather than only an average. This describes the evaluation direction; a published benchmark is not yet available.

The experiment is listed as testing. No reproducible benchmark dataset, hardware specification, or measured load results are documented in this portfolio yet.

Publish the benchmark conditions and query cases alongside results. Use those results to decide whether indexing, query design, or workload limits offer the most useful improvement for local retrieval.