Vector Store
Vector Store provides managed storage, indexing, querying, and retrieval for high-dimensional vector data. Use it for semantic search, retrieval-augmented generation (RAG), recommendation systems, and other workloads that depend on vector similarity search.
Introduction
With an API key, applications can connect to Vector Store and build context-aware systems that rely on vector similarity search. Vector Store can persist embeddings generated by SITE Cloud Inference APIs or other sources while keeping data in the SITE Cloud environment.
Overview
| Capability | Description |
|---|---|
| Store and manage vectors | Ingest vector data and metadata into a managed store optimized for similarity search. |
| Perform similarity search | Search across vectors to find results most relevant to an input query. |
| Integrate with AI Platform products | Use Embedding Inference and Document Parsing with Vector Store for retrieval workflows. |
A typical RAG workflow parses documents, embeds the extracted text, stores the embeddings in Vector Store, and retrieves relevant content for downstream generation.
Getting Started
- In Cloud Portal, navigate to Vector Store.
- Click Create.
- Select the tenant and business group.
- Enter the Vector Store name.
- Optionally add a description.
- Enter the vector dimensions that match the output of the embedding model you intend to use.
- Create the Vector Store.
- Copy the generated API key after the Vector Store is created.
Embedding model alignment
Use a single embedding model for each Vector Store. Mixing embedding models can reduce similarity search performance.
Default API key
A default Read & Write API key is created by default. Additional Read or Read & Write keys can be created from the Vector Store details page.
Creating Additional API Keys
Create additional API keys when you need different permission levels for different applications or workloads.
- Open the Vector Store details page.
- Go to the API Keys tab.
- Click Create.
- Enter the API key name.
- Optionally add a description.
- Choose a permission level.
- Click Create.
- Copy the generated key.
Allows read-only access, such as search or retrieval operations.
Allows read and write access, such as inserting, updating, or deleting vectors.
API key deletion
Deleted API keys cannot be recovered. Applications or workloads using a deleted key stop working.
Limits
| Limit | Default |
|---|---|
| Records per Vector Store instance | 500,000 |
Custom limits can be configured per instance for larger-scale or specialized use cases. Contact your designated Service Delivery Manager for more information.
Available Endpoints
Vector Store API endpoints are available in the AI Platform API documentation:
Endpoint and documentation access
The Vector Store endpoints and documenation mentioned above, are accessible only with an active connection to the SITE Cloud environment, and are not accessible from the public internet.
Glossary
| Term | Description |
|---|---|
| Vector | A numerical representation of data used for similarity comparisons. |
| Embedding | A dense vector representation that captures semantic meaning. |
| Similarity Search | Finding vectors that are most similar to a query vector. |
| RAG | Retrieval-Augmented Generation, a workflow that retrieves relevant information before generation. |
| API Key | A secure access token used to authenticate requests. |
| Endpoint | The URL where inference or vector operations are executed. |
| Permission | The access level granted to an API key. |