Storage Providers

SmartRAG storage provider configuration - Qdrant, Redis and InMemory storage options

Storage Provider Configuration

SmartRAG supports various storage providers:

Qdrant (Vector Database)

Qdrant is a high-performance vector database designed for production use with millions of vectors:

{
  "Storage": {
    "Qdrant": {
      "Host": "localhost",
      "UseHttps": false,
      "ApiKey": "",
      "CollectionName": "smartrag_documents",
      "VectorSize": 768,
      "DistanceMetric": "Cosine"
    }
  }
}

builder.Services.AddSmartRag(configuration, options =>
{
    options.StorageProvider = StorageProvider.Qdrant;
});

Advantages:

  • 🚀 High-performance vector search
  • 📈 Scalable (millions of vectors)
  • 🔍 Advanced filtering and metadata support
  • 🏢 Ideal for production

Disadvantages:

  • 🐳 Requires Docker or cloud service
  • 💾 Additional resource usage
  • 🔧 Setup complexity

Redis (High-Performance Cache)

Redis provides fast in-memory storage with vector similarity search capabilities using RediSearch:

{
  "Storage": {
    "Redis": {
      "ConnectionString": "localhost:6379",
      "Password": "",
      "Username": "",
      "Database": 0,
      "KeyPrefix": "smartrag:local:",
      "ConnectionTimeout": 30,
      "EnableSsl": false,
      "RetryCount": 3,
      "RetryDelay": 1000,
      "EnableVectorSearch": true,
      "VectorIndexAlgorithm": "HNSW",
      "DistanceMetric": "COSINE",
      "VectorDimension": 768,
      "VectorIndexName": "smartrag_vector_idx"
    }
  }
}

builder.Services.AddSmartRag(configuration, options =>
{
    options.StorageProvider = StorageProvider.Redis;
});

Advantages:

  • ⚡ Very fast access
  • 🔄 Automatic expire support
  • 📊 Rich data types
  • 🔍 Vector similarity search with RediSearch
  • 🏢 Suitable for production

Disadvantages:

  • 💾 RAM-based (limited capacity)
  • 🔧 Redis with RediSearch module required for vector search
  • 💰 Additional cost

RediSearch Module Required

Vector search requires RediSearch module. Use redis/redis-stack-server:latest Docker image or install RediSearch module on your Redis server. Without RediSearch, only text search will work (no vector similarity search).

Docker example:

docker run -d -p 6379:6379 redis/redis-stack-server:latest

InMemory (RAM Storage)

InMemory storage is ideal for testing and development, storing all data in RAM:

{
  "Storage": {
    "InMemory": {
      "MaxDocuments": 1000
    }
  }
}

builder.Services.AddSmartRag(configuration, options =>
{
    options.StorageProvider = StorageProvider.InMemory;
});

Use Cases:

  • 🧪 Testing and development
  • 🚀 Prototyping
  • 📊 Temporary data
  • 🔬 Proof of concept

Important

InMemory storage loses all data when application restarts. Not suitable for production!

Storage Provider Comparison

Compare storage providers to choose the best option for your use case:

Provider Performance Scalability Setup Cost Production Ready
Qdrant ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐⭐ ⭐⭐⭐ ⭐⭐⭐ Excellent
Redis ⭐⭐⭐⭐⭐ ⭐⭐⭐ ⭐⭐⭐ ⭐⭐⭐ Good
InMemory ⭐⭐⭐⭐⭐ ⭐ ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐⭐ Test only

Development and Testing

// For fast development and testing
options.StorageProvider = StorageProvider.InMemory;

Medium-Scale Applications

// Fast and scalable with RediSearch
options.StorageProvider = StorageProvider.Redis;

Large-Scale Production Applications

// Maximum performance and scalability for millions of vectors
options.StorageProvider = StorageProvider.Qdrant;

Next Steps

Database Configuration

Multi-database connections and schema analysis

Database Configuration

Audio & OCR

Whisper.net and Tesseract OCR

Audio & OCR