This guide covers Venice's APIs for applications that need to read and write data.
Venice supports multiple ingestion patterns depending on your data freshness requirements:
- Batch Push - Full dataset replacement from Hadoop/Spark with zero-downtime version swaps
- Stream Processor - Real-time updates via Apache Samza (requires hybrid stores)
- Online Producer - Direct writes from online services with durability guarantees
See Write APIs Overview for details.
Venice provides three client types with different latency/resource trade-offs:
| Client | Latency | Network Hops | Use Case |
|---|---|---|---|
| Thin Client | < 10ms | 2 (via Router) | Simple integration |
| Fast Client | < 2ms | 1 (direct) | Low latency |
| Da Vinci Client | < 1ms | 0 (local) | Ultra-low latency |
See Read APIs Overview for guidance on choosing a client.
For foundational concepts and deployment guides, see the Getting Started section.