feast-gridgain 1.0.0

This package makes it possible to use Ignite as online store for Feast.

feast-gridgain

This package enables the use of Apache Ignite or GridGain as an online store for Feast, providing high-performance, in-memory data storage and retrieval for feature serving.

Table of Contents

1. Features
2. Setup Instructions
   • Prerequisites
   • Installation
3. API Reference
4. Documentation
5. Example

Features

• Ignite/GridGain Integration: Leverages Ignite'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.

Setup Instructions

Prerequisites

• Python 3.11.7
• Running Apache Ignite or GridGain cluster

Installation

Install the package using pip:

pip install feast-gridgain

API Reference

IgniteOnlineStore / GridGainOnlineStore

Both classes implement 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.

For detailed information on these methods, refer to the docstrings in the source code.

Documentation

For an up-to-date documentation, see the GridGain Docs.

Example

For a comprehensive, real-world example of how to use this package, please refer to the following GitHub repository:

CGM Ignite Feast Kafka Example

This repository 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.