> For the complete documentation index, see [llms.txt](https://mariadb.com/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://mariadb.com/docs/gridgain/gridgain-8/integrations/ai/feast.md).

# Feast

Use GridGain 8 as an online store for Feast to serve features with high-performance, in-memory storage and retrieval.

The GridGain Feast integration provides a way of using GridGain as online store for Feast, providing high-performance, in-memory data storage and retrieval for feature serving.

## Features

* **GridGain Integration:** Leverages GridGain's in-memory database to provide online features for real-time model predictions.
* **Feature Management:** Feast manages feature definitions, versioning, and the synchronization between online and offline stores.

## Project Structure

The project consists of two main components:

1. **Ignite Online Store** (`online_store.py`): Sets up Apache Ignite as the online feature store.
2. **GridGain Online Store** (`gridgain_online_store.py`): Configures GridGain as the online feature store.

Both implementations provide similar functionality but are tailored to their respective systems.

## Prerequisites

The following is required to use the Feast integration:

* A running GridGain 8 cluster with an appropriate license, which the GridGain online store connects to.
* A Python version supported by the `feast-gridgain` package.

## Installation

Install the package using pip:

```bash
pip install feast-gridgain
```

## API Reference

### Online Store Reference

Both Apache Ignite and GridGain stores provide the following methods:

* `online_read(config, table, entity_keys, requested_features)` - Reads feature values from the online store.
* `online_write_batch(config, table, data, progress)` - Writes a batch of feature data to the online store.
* `update(config, tables_to_delete, tables_to_keep, entities_to_delete, entities_to_keep, partial)` - Updates the online store based on changes to the feature repository.
* `teardown(config, tables, entities)` - Cleans up the online store.

## Feast Tutorial

GridGain provides a comprehensive example of using Ignite for creating an online store for a Feast integration with Kafka. This tutorial provides a detailed implementation that demonstrates the integration of Ignite Online Store with Feast in a Continuous Glucose Monitoring (CGM) use case. It includes examples of configuration, feature definitions, and usage in different environments.

The full tutorial is available in the [How to Build a Machine Learning Feature Store With GridGain and Feast](/docs/gridgain/gridgain-8/gridgain8-get-started/how-to-guides/feast-feature-store-tutorial.md) page.