Qdrant Lite — Vector Database

High-performance vector database for AI applications. Store, search, and manage embeddings at scale with a single Rust binary.

Deploy and Host

Host your own Qdrant instance on Railway. This template provisions the Qdrant vector database with persistent storage for embeddings and metadata.

Why Deploy

Qdrant is the open-source vector database powering semantic search, recommendation systems, and RAG pipelines. Running it on Railway gives you a durable, auto-updating, globally reachable instance with:

Persistent embeddings — vector data survives restarts on a Railway volume
REST + gRPC APIs — integrate with any language or framework
Single binary — no external dependencies, fast startup
Tunable memory — scale RAM from 512MB to 1.5GB based on your dataset size
Production-ready — used by enterprises for billion-scale vector search

Common Use Cases

Semantic search — build search engines that understand meaning, not just keywords
RAG pipelines — power retrieval-augmented generation with your own embeddings
Recommendation systems — find similar products, content, or users
AI agents — give your LLM agents long-term memory via vector storage
Image search — index and search images by visual similarity

Deployment Dependencies

The deploy form pre-fills all required variables. No additional services needed — Qdrant runs as a single container with a volume for persistent storage.

After the first successful deploy:

Open https:///dashboard to verify the server is running
Test the API: curl https:///collections

About Hosting

Qdrant runs as a single container on Railway. Vector data persists on a Railway volume at /qdrant/storage. The server listens on port 6333 (REST API) and 6334 (gRPC API).

Features

Vector search — cosine, dot product, and Euclidean distance metrics
Payload filtering — attach metadata to vectors and filter by any field
REST + gRPC APIs — client libraries for Python, JavaScript, Rust, Go, and more
Quantization — reduce memory usage with scalar or product quantization
Hybrid search — combine dense and sparse vectors for best results
Distributed mode — scale horizontally with clustering (self-managed)
Single binary — no external dependencies, minimal resource footprint

Dependencies for

No external services required — Qdrant runs standalone
Optional: Redis companion — for distributed caching across multiple instances

Configuration

Variable	Description	Default
`PORT`	REST API port. Railway maps this to the public domain.	`6333`
`QDRANT__STORAGE__STORAGE_PATH`	Path to store vector data on the volume.	`/qdrant/storage`
`QDRANT__STORAGE__SNAPSHOT_PATH`	Path to store snapshots.	`/qdrant/storage/snapshots`
`QDRANT__SERVICE__HOST`	Bind address for the Qdrant server.	`0.0.0.0`
`QDRANT__SERVICE__HTTP_PORT`	REST API port.	`6333`
`QDRANT__SERVICE__GRPC_PORT`	gRPC API port.	`6334`
`QDRANT__LOG_LEVEL`	Logging level. Options: TRACE, DEBUG, INFO, WARN, ERROR.	`INFO`

Quick Start

After deployment, the Qdrant server is ready at your Railway public domain:

# Check server health
curl https://your-domain.up.railway.app/

# List collections
curl https://your-domain.up.railway.app/collections

# Create a collection
curl -X PUT https://your-domain.up.railway.app/collections/my_collection \
  -H 'Content-Type: application/json' \
  -d '{"vectors": {"size": 384, "distance": "Cosine"}}'

# Insert a point
curl -X PUT https://your-domain.up.railway.app/collections/my_collection/points \
  -H 'Content-Type: application/json' \
  -d '{"points": [{"id": 1, "vector": [0.1, 0.2, ...], "payload": {"text": "hello"}}]}'

# Search
curl -X POST https://your-domain.up.railway.app/collections/my_collection/points/search \
  -H 'Content-Type: application/json' \
  -d '{"vector": [0.1, 0.2, ...], "limit": 10}'

License

Qdrant is licensed under the Apache License 2.0. See the Qdrant GitHub repository for details.