Deploying Chronon Entities

Deploying Chronon Entities#

Once you have authored and tested a Chronon entity, such as a GroupBy, Join or StagingQuery, then all you need to do to productionize it is merge it into your repository. From there, the requisite DAGs and tasks will automatically get created to compute feature values and keep them up to date. See the Orchestration documentation for how to set this up.

GroupBy#

Once merged into production, GroupBys will get the following tasks and dags:

  1. (if online=True) A task in the {team_name}_chronon_batch DAG for daily batch uploads to the KV store.

  2. (if online=True and streaming) A task in the {team_name}_chronon_realtime DAG for a streaming job that writes to the KV store.

  3. (if backfill_start_date is set) A task in the {team_name}_chronon_batch DAG to front-fill the GroupBy snapshot table.

Join#

Once merged into production, Joins will get the following tasks and dags:

  1. (if online=True) Then the metadata for this join will get updloaded to the KV store as part of the chronon_metadata_upload DAG (all joins share the same task).

  2. (if check_consistency=True) A DAG gets created with the name pattern: online_offline_comparison_<team_name>_<join_name> which writes out daily consistency metrics as new upstream data lands.

StagingQuery#

Once merged into production, StagingQuerys will get the following tasks and dags:

  1. A task in the {team_name}_chronon_staging_query DAG for daily computation of the staging query.

Contents