# DIM table creation(SCD2), source selection

**URL:** <https://discourse.getdbt.com/t/dim-table-creation-scd2-source-selection/8368>\
**Category:** Help\
**Tags:** best-practice, snowflake\
**Created:** [May 20, 2023, 12:39am UTC](https://discourse.getdbt.com/t/dim-table-creation-scd2-source-selection/8368 "2023-05-20T00:39:25Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Borjoyzee](https://avatars.discourse-cdn.com/v4/letter/b/a3d4f5/32.png) [@Borjoyzee](https://discourse.getdbt.com/u/Borjoyzee)\
**Post date:** [May 20, 2023, 12:39am UTC](https://discourse.getdbt.com/t/dim-table-creation-scd2-source-selection/8368/1 "2023-05-20T00:39:26Z")

</div>

Hi guys,

I am completely new to dbt and there are few things a cant get my head around.

In simple: dbt can act as a substitute for datapipeline created in Snowflake using streams/tasks etc.?

Imagine I have source, each day this system extract data to S3. My goal is to created dimensional star schema.

In Snowflake I have external tables build on top of S3, So for example customer table will look like intial\_load/delta\_day1/delta\_day2 and so on. So it will be growing each day.

What should by my source for DIM tables? I get using snapshots. So I have also streams build on top of these external tables. And from these streams I am building snapshots.

So these stream I am using as SOURCES in dbt (for snapshots and snapshots for models down the line). Is this correct approach? Or should I use as source external tables? Or something different? What if something bad happend a source system will dump new data before snapshot creation? I will then have in stream 2 days, because stream wont be consumed. How can I how can I handle correct sequence loading into snapshot. So my history will be correct? What if I have to go back in time and regenerate the last three days?

Thank you