Pipelines & Streams
Batch and real-time workloads are designed on the same visual canvas. The platform picks the processing engine for your data size and topology: instant on small data, massively parallel at scale. You design the flow and never think about infrastructure.
- 184 ready-made building blocks: A ready component for every step from source to sink; add SQL or code blocks when custom logic is needed.
- Writes that never double or drop: Exactly-once, crash-safe writes by design: retries never produce duplicate records.
- Zero-downtime stream deploys: Blue/green deployment updates your stream topology without stopping live traffic.
- Schema management and replay: Schema compatibility is enforced before production; events can be replayed when needed.
Tip: Run the same pipeline on a small sample first; engine selection and the cost estimate update automatically.