{% hint style="danger" %} We strongly encourage all users to upgrade from Feast 0.9 to Feast 0.10+. Please see this for an explanation of the differences between the two versions. A guide to upgrading can be found here. {% endhint %}
- Added Feast Job Service for management of ingestion and retrieval jobs
- Added support for Spark on K8s Operator as Spark job launcher
- Added Azure deployment and storage support (#1241)
Note: Please see discussion thread above for functionality that did not make this release.
- Add support for AWS (data sources and deployment)
- Add support for local deployment
- Add support for Spark based ingestion
- Add support for Spark based historical retrieval
- Move job management functionality to SDK
- Remove Apache Beam based ingestion
- Allow direct ingestion from batch sources that does not pass through stream
- Remove Feast Historical Serving abstraction to allow direct access from Feast SDK to data sources for retrieval
- Label based Ingestion Job selector for Job Controller #903
- Authentication Support for Java & Go SDKs #971
- Automatically Restart Ingestion Jobs on Upgrade #949
- Structured Audit Logging #891
- Request Response Logging support via Fluentd #961
- Feast Core Rest Endpoints #878
- Improved integration testing framework #886
- Rectify all flaky batch tests #953, #982
- Decouple job management from Feast Core #951
- Batch statistics and validation #612
- Authentication and authorization #554
- Online feature and entity status metadata #658
- Improved searching and filtering of features and entities
- Python support for labels #663
- Improved job life cycle management #761
- Compute and write metrics for rows prior to store writes #763
- Streaming statistics and validation (M1 from Feature Validation RFC)
- Support for Redis Clusters (#478, #502)
- Add feature and feature set labels, i.e. key/value registry metadata (#463)
- Job management API (#302)
- Clean up and document all configuration options (#525)
- Externalize storage interfaces (#402)
- Reduce memory usage in Redis (#515)
- Support for handling out of order ingestion (#273)
- Remove feature versions and enable automatic data migration (#386) (#462)
- Tracking of batch ingestion by with dataset_id/job_id (#461)
- Write Beam metrics after ingestion to store (not prior) (#489)