[FEATURE]: Enable Liquid Clustering for tables migrated

[artsheiko]

Is there an existing issue for this?

• I have searched the existing issues

Problem statement

UC enablement requires recreating tables for managed ones. It could be great to allow LC enablement for all / selected tables.

Proposed Solution

Once CLUSTER BY AUTO is shipped, we could mark tables in mapping.csv as to-be-activated-for-LC by default. Using the file, teams could decide whether to enable LC or do nothing (meaning they plan to continue to leverage their existing maintenance operations).

As an extension, we could allow the manual inclusion of a list of columns to be used as keys for LC for one or another table.

Finally, we could identify what is used for ZORDER, re-use the same columns in LC (if any), or invite to overwrite a list with new ones, or just set AUTO.

Additional Context

It seems at first ANALYZE will become available for PO (blog). As of 2024-12-09, the feature is in Gated Public Preview.
Next, Liquid will allow for a dynamic selection of clustering keys (CLUSTER BY AUTO), based on the Q4 roadmap, Public Preview should become available by ~January.

• Once available, Choose clustering keys section from the documentation will be relevant when the customer knows what exactly should be used as keys, but I believe auto-selection will become more and more useful option.
• Given Predictive Optimisation supports LC, CLUSTER BY AUTO for migrated tables could allow for a better UX compared to ZORDER from the day 0.