Model the Snowflake access history

[ian-whitestone]

We need to model the snowflake.account_usage.access_history, since it contains a bunch of nested JSON columns that aren't easy to query. We need to be able to answer questions like:

1. most expensive queries against a particular table with poor pruning (help inform clustering keys)
2. most frequently executed queries against a table (help inform clustering keys)
3. tables that haven't been accessed in last X days

Bunch of good work in here for reference: https://github.com/jaysobel/dbt-snowflake-queries. Need to connect with Jay, he had found some gotchas with these models.

[ian-whitestone]

a really cool extension of this would be to use an AST parser or SQL fluff to include columns like:

• array of columns used in join condition
• array of columns used in where clause

This would be useful for providing clustering key recommendations