https://dagster.io/ logo
#ask-community
Title
# ask-community
i

Ivan Tsarev

05/11/2023, 2:02 PM
Hi folks! I have pretty broad-looking for inspiration type of question. I have dbt python model which requires Dataproc cluster to be calculated. So I want to orchestrate cluster creation and dbt Cloud run with dagster. I`m already have ops to spin up cluster and playing with dbt Cloud connection. But how should I compose these two tasks in terms of Dagster abstraction? Cluster management is ops and probably should not become assets, since there is no data materialisation behind it. Dbt cloud on the other hand operates with models as Assets.| And I cannot find an easy way to define dependency of software-define graph to completely different ops. So how should I approach this issue? The easiest way I see, is create/import separate ops to start dbt jobs without any asset importing and stuff. It feels kinda natural from my airflow experience, where you have "just" tasks, but I'm not quite sure what I will lose if I'm not gonna use assets at all since it is quite a cornerstone of Dagster?