Interested to know what approach one would use with Snowflake when you have multiple sources arriving at different times of the day that you want to process as they come in.
Teradata would allow you to create a view that could be inserted into so no problem, but without being able to create views to alias I dont see how to get around this.
THis is a very common scenarios in corporate data warehouses, account balance tables for example that get load from many disparate sources.
Will DBT move away from the one table one transform model soon? After all conforming data from mulitple sources is sort of the point of data analytics in the first place.
I’m unfamiliar with Teradata so this could be ignorance on my part, but generally you don’t insert into views; do they use different terminology? I also don’t know what you mean by “create views to alias”, could you clarify those points?
dbt supports (and strongly encourages!) taking multiple source tables and turning them into a single transformed mart table. It doesn’t support building multiple output tables from a single model file, but I don’t think that’s what you’re referring to.
If you use a traditional view where the underlying tables are accessed live on every query, then dbt would work great for that. If you want to materialize them as a table, then your data would obviously only be as up to date as the last
dbt run which built that table. (Keep in mind that if you have a very very large table, you can use incremental models to only process the new data).