dbt-external-tables VS dbt-synapse

You can setup your CDC process such that it will create and evolve the tables for you, e.g. by using jdbc connector. So the moment your OLTP database schema changes, it gets picked up by the CDC process and it will propagate it to your OLAP database. If you want to have more control over schema evolution, you can also do it in dbt using dbt-external-tables package.

This looks like what is done by https://github.com/dbt-labs/dbt-external-tables/blob/main/sample_sources/snowflake.yml right ?

If you are familiar with dbt, it should be straightforward to do this using dbt and dbt-external-tables.

We keep track of the schema in git using dbt-external-tables. If a new column is added upstream, whoever made the change is responsible for updating the schema in dbt. It's just a few lines of code. Once PR is merged, the external table that reads the S3 files is refreshed. We never remove columns.

In addition to the built-in seed command you can also install this package called dbt-external-tables which will automate the creation of pointers/views to external data sources living on S3/GCS/Azure Blob Storage for several data warehouses. Then you can simply select from these external sources when building your models.

For the external tables, I'm using the following plugin: https://github.com/fishtown-analytics/dbt-external-tables

I'm using dbt-synapse: https://github.com/dbt-msft/dbt-synapse

Have you tried dbt-synapse? I'm the maintainer and I'd love to hear what you think is missing. Also, w.r.t. dbt cloud support for Synapse -- it's something we're working on, but we're we need buy-in from MSFT first!