search

dbt-labs / docs.getdbt.com

The code behind docs.getdbt.com
https://docs.getdbt.com/
Apache License 2.0
118 stars 943 forks source link

[Core] what are "pure marts"? #6329

Open AlexVialaBellander opened 5 days ago

AlexVialaBellander commented 5 days ago

Link to the page(s) on docs.getdbt.com requiring updates

https://docs.getdbt.com/best-practices/how-we-structure/4-marts

Tell us more about this update

What do you mean with "pure marts" here? A colleague asked me and I cant explain it well enough. Could it be expanded?

Reviewers/Stakeholders/SMEs

Not sure who that would be, but probably the authors of the dbt docs. it's your opinionated guide so yea?

Related GitHub issues

No response

Additional information

No response

runleonarun commented 1 day ago

Thank you for opening this issue! @dbeatty10 can you expand on pure marts or recommend another SME?

dbeatty10 commented 1 day ago

Current docs

image

Explanation

This is more simple than it looks! In this context, "pure marts" just means marts that do not include any time-based rollups like orders_per_day or orders_per_week, but instead use the semantic layer for dynamic and flexible downstream aggregations.

Suggestion

To avoid confusion, we could update our docs to be something like this:

✅ Focus on entity naming. Use plain English to name the file based on the concept that forms the grain of the mart customers, orders. It's usually best to avoid including rollups on time dimensions in this layer (such as orders_per_day). Instead, consider refactoring such rollups into the semantic layer, enhancing flexibility and scalability.