[Column-level encoding] Add support for column-level encoding to Redshift adapter

[gordonr]

Is this your first time submitting a feature request?

• [X] I have read the expectations for open source contributors
• [X] I have searched the existing issues, and I could not find an existing issue for this feature
• [X] I am requesting a straightforward extension of existing dbt-redshift functionality, rather than a Big Idea better suited to a discussion

Describe the feature

Currently the Redshift adapter does not support column-level encoding, which is used to configure compression. This appears to be supported in the Redshift CREATE TABLE statement, as documented here. An approach might be to translate a Redshift-specific dbt config setting to the appropriate CREATE TABLE statement modifications to achieve the desired outomce.

Describe alternatives you've considered

If a dbt model is an incremental view, it would be possible to manually modify column-level encoding after the initial build of the Redshift table, but those changes would be lost if a full refresh were done.

Who will this benefit?

This would theoretically benefit all Redshift customers.

Are you interested in contributing this feature?

No response

Anything else?

No response

[TuringND]

For what is worth, one of the values of having column-level encoding relates to incremental models. Currently compression post-hook makes a deep-copy of the model.

This strategy, though sound for table materializations, antagonizes with incremental materializations as it derives in a full copy of the data. When the model data is big enough, using incremental provides a substantial increase in performance, performance that is lost due to the time spent in the deep-copy.

I was going to post about this in a separate git issue / create a topic in discourse but this feels more appropriate.