Question about how Redshift Destination works

[morA123R]

Hi,
I wanted to ask a few questions about the Redshift destination, and hopefully I can get some answers about it:

1. I have several sources that I want to connect to the same redshift destination, and sync them in an incremental append+deduped mode. Does each connection that I create will open a new connection to Redshift, or will it use the same connection? I would like to understand if I need to handle the rate limit of concurrent open connections.
2. I saw that Airbyte automatically creates a destination table in Redshift according to the source schema, and I saw that it defined the SORTKEY to ( _airbyte_unique_key, _airbyte_emitted_at ). If I want to change it to AUTO for better optimization, since I also query the data, could it affect the way Airbyte works?
3. If I have a source that is scheduled to sync to Redshift every hour, and the first increment takes 2 hours, and the second increment takes 30min so it can finish before the first one. Will I still get the right order of the increments in the destination table?

Thanks in advance.

[marcosmarxm]

1. Each Airbyte connection will open a new connection in Redshift.
2. The SORTKEY is used by normalization which is a deprecated module used by destinations. You can extract the dbt project and change as your will, see here.
3. Airbyte will have only one sync running at the time. If the first didn't finish Airbyte will schedule but not trigger the next sync.

[morA123R]

Thanks @marcosmarxm for the explanation.

1. So if I want to handle the rate limit of concurrent open connections to the redshift, can I use the MAX_SYNC_WORKERS configuration to achieve that? If so, where can I change this configuration?
2. Is the SORTKEY not used for syncing the increments as well? And regarding the custom dbt project, can you elaborate on it? Can I use it to tell Airbye how to create the destination table, instead of what Airbyte automatically creates? which means I will be able to change the SORTKEY and also some of the data types if I want.
   But if it runs in every sync, will it rerun the creation table sql script that I wanted to change, and can override the destination table?
   Does the dbt project is used only for data transformation using sql, or can it run also a python code, for example?
3. Great, thank you.