Table sources are used directly via a view. Featureform will never write to a primary source.
Files are copied into a Snowflake table via a Kubernetes Job kicked off by our coordinator. If scheduling is set, the table will atomically be re-copied over.
SQL transformations are used to create a view. By default, those views are materialized and updated according to the schedule parameter. Deprecated transformations are converted to un-materialized views to save storage space.
When a feature is registered, Featureform creates an internal transformation to get the newest value of every feature and its associated entity. A Kubernetes job is then kicked off to sync this up with the Inference store.
Every registered feature and label is associated with a view table. That view contains three columns, the entity, value, and timestamp. When a training set is registered, it is created as a materialized view via a JOIN on the corresponding label and feature views.
First we have to add a declarative Snowflake configuration in Python.
import featureform as ff
name = "snowflake_docs",
description = "Example inference store",
team = "Featureform",
username = snowflake_username,
Once our config file is complete, we can apply it to our Featureform deployment
featureform apply snowflake_config.py --host $FEATUREFORM_HOST