## Vector Search on Object Storage. Scale Without the RAM Tax.

Most vector databases keep everything in RAM. LanceDB stores data in object storage. Quantized indexes fit in memory. Full-fidelity vectors fetched from storage for reranking. Memory-like search performance, object storage cost.

## The Power of the Lance Format

##### Vector Search
- Fast scans and random access from the same table — no tradeoff
- Zero-copy access for high throughput without serialization overhead

##### Multi-Modal
- Raw data, embeddings, and metadata in one table — not pointers to blob storage
- No separate metadata store to keep in sync

## Comparison

| **Feature** | **Legacy Vector Database** | **LanceDB** |
| ----------- | -------------------------- | ------------ |
| **Cost** | RAM-bound. $3-5/GB/month at scale. | Object storage. $0.02/GB/month. |
| **Scale** | Limited by RAM. | 20 PB largest table. 20K+ QPS. |
| **Search** | Vector search. Full-text via integration. | Vector, full-text, SQL in one query. |
| **Data model** | Embeddings only. Raw data elsewhere. | Search, analytics, feature engineering, training. |
| **Purpose** | HNSW graph mutation. Slow writes. | IVF partitions. Writes don't block reads. |
| **Best for** | Small, static datasets. | Production workloads at scale.

## Enterprise-Grade Requirements

### Security

Granular RBAC, SSO integration, and VPC deployment options.

### Governance

Data versioning and time-travel capabilities for auditability.

### Support

Dedicated technical account management and guaranteed SLAs.