How will/does dbt-server handle multiple, concurrent requests? Is there a plan to make dbt-server stateless and independent of the presence of a manifest?

[vladimirdinolov-8451]

We are using dbt-server and have even made some adjustments to the source code to solve for things like ignoring the parsing of a manifest to enable dynamic models. Will dbt-server eventually have the ability to "stream" vs. "batch" data? Another way to phrase this, will dbt-server continue to be stateful or is there a plan to make it stateless?

What do I mean by stateful vs stateless? Specifically, understanding if dbt-server will be able to run regardless of the presence of a manifest and the need to parse that manifest. My current understanding is that dbt-server is stateful and references the manifest to communicate the latest state.

If this doesn't 100% make sense, please let me know as I'd love to flesh out the answer to this. Thank you for all your hard work!