Databricks: table hygiene

[tasozgurcem11]

Summary

Define whether Databricks needs a table-hygiene review check (analogous to Redshift SVV_TABLE_INFO / BigQuery unpartitioned tables).

Context

Redshift reports unsorted regions, stale stats, and skew via table_hygiene. BigQuery flags large unpartitioned tables. Databricks in #3 returns {} from table_hygiene — no Databricks-specific hygiene verdict exists yet.

Open questions

• Does Unity Catalog or Databricks system metadata expose actionable maintenance signals (e.g. Z-ordering, optimize stats, file layout) comparable to Redshift/BigQuery?
• Is the check valuable enough to justify another warehouse-specific code path, or is "not applicable" the right long-term answer?

Proposed approach

1. Research available UC / system table signals and grants required.
2. Decide: implement a verdict, or document explicitly that Databricks has no hygiene check (like Snowflake for partitioning).
3. If implemented: SQL builder, client method, scorecard section, tests, docs.

Related

• PR feat: add Databricks support for dbt-debt scan #3