[FEATURE]: Migrate tables in unsupported filesystem

[qziyuan]

Is there an existing issue for this?

• I have searched the existing issues

Problem statement

External tables stored in adl:// and wasbs:// will be crawled and marked with What.EXTERNAL_NO_SYNC.
We will need more What enum to differentiate following scenarios:

• Hiveserde tables, like ParquetHiveSerDe, which cannot be SYNC, but can be in place migrated by creating a UC table with supported data source (for example create external table ... using parquet ... location)
• Hiveserde tables that have to be migrated using CTAS
• Tables in unsupported filesystem like adl:// and wasbs://. It require either:
  • migrate the storage to ADLS Gen2 first and update the HMS table location, then migrate to UC.
  • or deep clone or CTAS the table to a UC.

Proposed Solution

• Add more What enum.
• Discuss the strategy of how to migrate those tables in the future.

Additional Context

Related issue:

• Report on ADLSv1 and Azure Blob Storage as EXTERNAL table sources don't work with UC #355 which reports unsupported table in dashboard.
• [BUG]: Migrate UC External Location should skip unsupported filesystem #1064 Migrate UC External Location should skip unsupported filesystem

[nfx]

effort might be: 3weeks for cloud-level copy or few days for CTAS