Is your feature request related to a problem? Please describe.
- Users commonly want to leverage their existing tables created by dbt models within Feast. Right now, this requires manual creation
- When materializing features to the online store, often, it's desirable to have dbt run first to create the upstream table prior to materializing an interval. This requires manual work today.
Describe the solution you'd like
-
A mechanism to import dbt models (e.g. feast dbt-import) that lets users import from a directory of models, or configure a directory and enable individual models
-
The above mechanism should generate a data source + feature view with pre-populated tags to track the origin dbt model. Something like
driver_stats_source = BigQuerySource(
name="driver_hourly_stats_source",
table="AUTO-IMPORTED-NAME",
tags={
"feast.dev/dbt-model-path": "INSERT MODEL PATH"
"feast.dev/dbt-model-serialized": "<serialized dbt model to be shown in UI>"
}
)
Describe alternatives you've considered
Additional context
Add any other context or screenshots about the feature request here.
Is your feature request related to a problem? Please describe.
Describe the solution you'd like
A mechanism to import dbt models (e.g.
feast dbt-import) that lets users import from a directory of models, or configure a directory and enable individual modelsThe above mechanism should generate a data source + feature view with pre-populated tags to track the origin dbt model. Something like
Then these models could potentially be shown directly in the Feast Web UI too (there was prior art here e.g. https://github.com/feast-dev/feast/blob/v0.24-branch/ui/src/pages/data-sources/DataSourceDbt.tsx
Potentially, if configured, materialize can also run dbt run for the upstream model
Describe alternatives you've considered
Additional context
Add any other context or screenshots about the feature request here.