Back to articles
AI & Cloud

Optimizing Vector Databases for RAG: Comparing Pgvector, Qdrant, and Milvus on VPS

A guide to optimizing vector databases for RAG using Pgvector, Qdrant, and Milvus on VPS, including configuration, performance comparison, and practical deployment steps.

DPTCloud Engineering4 min read
Optimizing Vector Databases for RAG: Comparing Pgvector, Qdrant, and Milvus on VPS

Optimizing Vector Databases for RAG on VPS

Deploying Retrieval-Augmented Generation (RAG) systems is increasingly common in AI applications, especially when precise information retrieval from large datasets is required. Vector databases play a critical role in storing and querying vector embeddings. In this guide, we compare the performance of three popular vector database platforms: Pgvector, Qdrant, and Milvus, when deployed on a Virtual Private Server (VPS).

Performance Comparison Between Pgvector, Qdrant, and Milvus

We evaluate the three platforms based on query speed, storage efficiency, scalability, and stability on a VPS environment.

1. Pgvector

Pgvector is a PostgreSQL extension that enables direct storage and querying of vector data within a relational database. Its main advantage is seamless integration with existing database architectures, requiring no architectural changes.

To install Pgvector on a VPS, follow these steps:

# Install PostgreSQL
sudo apt update
sudo apt install postgresql postgresql-contrib

# Install Pgvector
cd /tmp
git clone https://github.com/pgvector/pgvector.git
cd pgvector
make && make install

# Enable the extension in PostgreSQL
sudo -u postgres psql -c "CREATE EXTENSION IF NOT EXISTS vector;"

2. Qdrant

Qdrant is a dedicated vector database designed for fast and efficient vector queries. It supports features such as partitioning, label-based filtering, and data filtering.

Qdrant can be easily deployed using Docker:

docker run -p 6333:6333 -p 6334:6334 -v /qdrant/data:/qdrant/data qdrant/qdrant

3. Milvus

Milvus is an open-source vector database designed for large-scale vector search tasks. It supports Approximate Nearest Neighbors (ANN) and offers excellent scalability.

Deploy Milvus on a VPS using Docker:

# Install Milvus with Docker
docker run -d --name milvus -p 19530:19530 -p 19121:19121 milvusdb/milvus:latest

Performance Comparison Table

Factor Pgvector Qdrant Milvus
Query Speed Average High Very High
Scalability Limited Average Very High
Stability High Average High
Integration with PostgreSQL Perfect None None

System Requirements and Optimization Configuration

To ensure optimal performance, optimize your VPS system:

  • RAM: At least 16GB RAM for heavy RAG tasks.
  • Storage: SSD or NVMe for faster read/write speeds.
  • Network: Stable, high-speed connection to reduce latency.

Testing and Verification Methods

To verify performance, use tools like ab (Apache Bench) or wrk for query testing. For example:

# Test Pgvector query performance
ab -n 1000 -c 10 http://localhost:5432/query

Troubleshooting

If you encounter deployment issues, check the following:

  • Ensure services (PostgreSQL, Qdrant, Milvus) are running.
  • Check logs for specific errors.
  • Verify system resources (RAM, CPU) are sufficient for the task.

References

DPTCloud Tier III Infrastructure

High-RAM Cloud VPS Plans Optimized for AI, Databases & Cache

Enterprise RAID 10 NVMe SSDs, CMC & FPT District 9 uplinks with Anycast Anti-DDoS and 99.99% SLA.

View 22 VPS Plans
ECO · vn-02Recommended

VPS-4E

297.000 ₫ / mo

10% VAT included

  • CPU: 4 vCPU
  • RAM: 8 GB
  • SSD: 50 GB NVMe
ECO · vn-02

VPS-5E

614.900 ₫ / mo

10% VAT included

  • CPU: 8 vCPU
  • RAM: 16 GB
  • SSD: 100 GB NVMe
ADVANCED · vn-03Recommended

VPS-4T

548.900 ₫ / mo

10% VAT included

  • CPU: 4 vCPU
  • RAM: 8 GB
  • SSD: 50 GB NVMe

Related Articles

View all
ZaloZalo SupportDiscord0384 753 478