# Fivetran and incremental loads

**URL:** <https://discourse.getdbt.com/t/fivetran-and-incremental-loads/2355>\
**Category:** Archive\
**Created:** [April 2, 2021, 7:34pm UTC](https://discourse.getdbt.com/t/fivetran-and-incremental-loads/2355 "2021-04-02T19:34:52Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![jomach](https://sea2.discourse-cdn.com/flex020/user_avatar/discourse.getdbt.com/jomach/32/686_2.png) [@jomach](https://discourse.getdbt.com/u/jomach)\
**Post date:** [April 2, 2021, 7:34pm UTC](https://discourse.getdbt.com/t/fivetran-and-incremental-loads/2355/1 "2021-04-02T19:34:52Z")

</div>

Hello Guys,

I’m thinking on using fivetran + dbt for my next Dwh project. I have one question, fivetran syncs the tables that I need to create My facts and dims tables. From what I know, usually we truncate the tables before loading the data from source system. This way we can really do a nice incremental load.

After all if I have a staging area from 1TB it does not make since to keep it forever there and reload all the facts and dim for multiple years if I just want to load 1 day.

Is there anyway approach to deal with this? Like delete the data from the staging area after it was inserted into the dim and fact tables

---

<div class="post-metadata">

**Author:** ![reeves](https://sea2.discourse-cdn.com/flex020/user_avatar/discourse.getdbt.com/reeves/32/687_2.png) [@reeves](https://discourse.getdbt.com/u/reeves)\
**Post date:** [April 3, 2021, 9:24pm UTC](https://discourse.getdbt.com/t/fivetran-and-incremental-loads/2355/2 "2021-04-03T21:24:38Z")

</div>

As systems get larger it would not be ideal to truncate and reload. What I think you should consider is an incremental load and there are things in Fivertran and dbt target databases that can help.

**Example:**

There is a column in Fivetrans managed tables (\_FIVETRAN\_SYNCED) that can help with what is new/updated.

[https://fivetran.com/docs/getting-started/system-columns-and-tables](https://fivetran.com/docs/getting-started/system-columns-and-tables)

There are also source database features like streams in Snowflake that can do the change detection as well.

[https://docs.snowflake.com/en/user-guide/streams.html](https://docs.snowflake.com/en/user-guide/streams.html)

hope this helps

---

<div class="post-metadata">

**Author:** ![jomach](https://sea2.discourse-cdn.com/flex020/user_avatar/discourse.getdbt.com/jomach/32/686_2.png) [@jomach](https://discourse.getdbt.com/u/jomach)\
**Post date:** [April 4, 2021, 6:05am UTC](https://discourse.getdbt.com/t/fivetran-and-incremental-loads/2355/3 "2021-04-04T06:05:48Z")

</div>

So we replicate the whole erp db into the dwh dB where dbt runs ? We keep this data forever there ?

Would be great to have an example how to deal with this …

What I would do now is:  
1- sync the source table with fivetran  
2 - create dim and fact tables with incremental loads  
3 - truncate the source table with a post hook

This way The staging area from the dwh is always clean. If I need a re-sync then Incan trigger it with fivetran

Ist this correct ?

Jorge