System tables are a Databricks‑hosted, read‑only analytical store shared to your workspace via Delta Sharing; they aren’t modifiable (no indexes you can add), and the first read can have extra overhead on a very small warehouse. This can make “simple” selects feel slow on X‑Small, especially if the warehouse was cold or the table had to be hydrated from the sharing layer.

If possible, run on a larger or Serverless warehouse for interactive reads; X‑Small is cost‑efficient but can have higher per‑query latency, particularly on first access to shared datasets and avoid SELECT *; project only the columns you need to minimise bytes scanned.