[IMPALA-8620] Add compression codecs as table/column properties

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: Backend
• Labels: compression parquet
• Epic Color: ghx-label-5
• Assignee: Abhishek Rawat
• Reporter: Abhishek Rawat
• Created: 04/Jun/19 11:09
• Updated: 04/Jun/19 11:09

Description

COMPRESSION_CODEC is supported as a query option.

It probably makes sense to provide a finer granularity control on the scope of compression such as table and/or even columns.

There could also be scenarios where a select few columns (such as variable length strings) could be compressed while others (such as integer family columns) remain uncompressed for performance. Parquet's unit of compression is a page and so it makes sense to provide capability to compress a given column.