Milvus Configuration Guide

June 27, 2025 ยท View on GitHub

Overview

This guide explains how to configure and connect to Milvus vector database in different deployment scenarios.

Environment Variables

Configure your Milvus connection using these environment variables:

  • MILVUS_ENDPOINT: The connection endpoint for your Milvus instance
  • MILVUS_TOKEN: (Optional) Authentication token for secure connections

Deployment Options

1. Milvus Lite (Local Development)

  • Best for: Development, testing, and small datasets
  • Setup: Set MILVUS_ENDPOINT to a local file path (e.g., ./milvus.db)
  • Features:
    • Milvus Lite is a vector database that runs directly in Python
    • Stores all data in a local file, ideal for quick prototyping
    • No additional infrastructure setup required

2. Milvus Server (Self-Hosted)

  • Best for: Large-scale production deployments
  • Setup:
    • Set MILVUS_ENDPOINT to your server URI (e.g., http://localhost:19530)
    • Optionally set MILVUS_TOKEN for authentication
  • Features:
    • Full functionality, scales to billion vectors
    • Supports Docker (Milvus Standalone) and Kubernetes (Milvus Distributed): installation guide

3. Zilliz Cloud (Fully Managed Milvus)

  • Best for: Production deployments with managed infrastructure
  • Setup:
    • Set MILVUS_ENDPOINT to your Zilliz Cloud public endpoint
    • Set MILVUS_TOKEN to your API key
  • Features:
    • Fully managed service available on Azure, AWS and Google Cloud
    • Zero DevOps required with automatic scaling and high availability
    • Start with a free trial

Additional Resources