Proposal: Add openGauss Database Support

[qifengleqifengle]

Description

This proposal adds native support for the openGauss database as a vector_io provider. With this enhancement, users can connect Llama Stack directly to an openGauss instance for vector storage and similarity search.

Motivation

Currently, Llama Stack does not support openGauss out of the box. openGauss is a high-performance, open-source relational database with a rapidly growing ecosystem. Adding first-class support for its vector capabilities offers the following benefits:

• Expands Database Choices: Increases the range of compatible vector stores, offering a robust, open-source option for users.
• Unified Data Strategy: Enables teams already using openGauss to store both their operational data and vector embeddings within the same system, simplifying architecture.
• Leverages Native Performance: Utilizes openGauss's underlying performance and scalability for vector operations without requiring external adapters.
• Engages a Broader Community: Encourages contributions and adoption from the growing openGauss user base.

Proposal

I have implemented an OpenGaussVectorIOAdapter module that:

• Integrates with the existing VectorIO provider interface.
• Handles direct connection to openGauss using psycopg2.
• Supports core vector store operations: register_vector_db, unregister_vector_db, insert_chunks, and query_chunks.
• Includes a full suite of unit and integration tests to ensure functionality and stability.

This implementation is designed to integrate seamlessly with the existing Llama Stack provider framework and introduces no breaking changes. I am prepared to open a Pull Request with the completed code and tests.

Inspiration

This proposal was motivated by the increasing adoption of openGauss in production environments and the opportunity to provide Llama Stack users with more powerful, integrated choices for their RAG and agent memory backends.