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 instanceMILVUS_TOKEN: (Optional) Authentication token for secure connections
Deployment Options
1. Milvus Lite (Local Development)
- Best for: Development, testing, and small datasets
- Setup: Set
MILVUS_ENDPOINTto 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_ENDPOINTto your server URI (e.g.,http://localhost:19530) - Optionally set
MILVUS_TOKENfor authentication
- Set
- 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_ENDPOINTto your Zilliz Cloud public endpoint - Set
MILVUS_TOKENto your API key
- Set
- 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
- Milvus Documentation: Comprehensive guides and API references
- Zilliz Cloud Documentation: Managed service documentation
- Milvus GitHub Repository: Source code and issues
- Milvus Community: Join our Discord community for support