Namespaces provide logical separation for different use cases, applications, or environments while maintaining independent configurations and security controls.

​
Quick Start Video

Watch this 5-minute live walkthrough to see namespaces in action. The video covers creating, managing, and implementing namespaces for multimodal data isolation.

​
Key Concepts

Isolation

Each namespace operates as a completely isolated environment with its own:

  • Resource allocation
  • Access controls
  • Configuration settings

Indexing

Configure vector and payload indexes for:

  • Feature extraction
  • Search capabilities
  • Data organization

Vector Versioning

Seamlessly manage embedding model transitions:

  • A/B test different embedding models
  • Batch re-embed existing data
  • Zero-downtime production cutover

​
Common Challenges We Solve

​
Embedding Model Evolution

Managing embedding model transitions can be complex. For example, when upgrading from OpenAI’s clip-vit-base-patch32 to Nomic AI’s nomic-embed-vision-v1.5, you face several challenges:

  • Re-embedding millions of existing images
  • Testing retrieval quality across different models
  • Maintaining service continuity during transitions
  • Managing multiple embedding versions simultaneously

Namespaces provide a elegant solution by:

  • Creating isolated environments for different embedding models
  • Enabling side-by-side quality comparisons
  • Supporting seamless production cutover
  • Switching between versions with a simple X-Namespace header

​
Vector Indexes and Embedding Models

Mixpeek provides managed vector indexes that automatically handle embedding updates and model improvements.

For an exhaustive list of available Embedding Models, check out Available Models

# Available vector indexes
embedding_models = {
    "image": {
        "modalities": ["image", "text"],
        "type": "dense",
        "size": 512
    },
    "multimodal": {
        "modalities": ["image", "video", "text"],
        "type": "dense", 
        "size": 1408
    },
    "text": {
        "modalities": ["text"],
        "type": "dense",
        "size": 1024
    },
    "keyword": {
        "modalities": ["text"],
        "type": "sparse"
    }
}

​
Payload Indexes

​
Best Practices

​
Organization Strategies

1

Environment Separation

Create separate namespaces for development, staging, and production

2

Use Case Separation

Isolate different applications like product search, content discovery, etc.

3

Client Separation

Maintain distinct namespaces for different customers or regions

​
Naming Conventions

Use descriptive, lowercase names
Include environment or purpose
Be consistent across teams
Add relevant prefixes/suffixes

Examples:

prod-ecommerce
dev-content-search
staging-recommendations

Learn more about namespace management in our API Reference.

Was this page helpful?

Suggest editsRaise issue