Available
CDC source
PostgreSQL
Logical replication / WALCapture ordered row changes from PostgreSQL and maintain resumable projection state.
- Snapshot bootstrap
- SQL recomposition
- Checkpointed
Capture database changes, consume durable broker events, and deliver joined projections or live subscriptions without application dual-writes.
Sources capture change. Brokers carry realtime events. Sinks materialize projections. Delivery protocols serve clients.
Capture ordered row changes from PostgreSQL and maintain resumable projection state.
Stream row mutations from MySQL-compatible databases using durable binlog positions.
Tail selected collections with resume tokens and optional snapshot bootstrap.
Turn graph changes and multi-hop Cypher projections into search-ready documents.
Consume raw records or unwrap Debezium envelopes from Kafka-compatible brokers.
Fan out low-latency events when replay and broker persistence are not required.
Deliver acknowledged events with broker cursors, replay, and reconnect recovery.
Use Redis stream IDs for resumable GraphQL and native WebSocket delivery.
Write fixed, relation-routed, or projection-targeted documents with external ordering.
Target compatible Elasticsearch APIs through the same tested bulk sink contract.
Expose typed realtime operations with broker-backed cursor resume semantics.
Deliver broker events directly to clients that do not need a GraphQL layer.
CDC and realtime delivery share the same engine and Fleet administration without conflating their reliability contracts.
Ordered source positions advance only after projection work and sink acknowledgement complete.
Explore CDCDurable providers preserve cursor-based replay while clients choose GraphQL or the native socket protocol.
Explore realtimeThese contracts are under consideration. They are not delivery commitments and do not have published release dates.
Write durable analytical files for downstream lakehouse and batch workflows.
Forward selected change events to customer-managed HTTP endpoints.
Move operational changes into analytical warehouse tables.
Use Kafka partitions and offsets as a realtime subscription transport.
Support ephemeral Redis channels where stream persistence is unnecessary.
Configure one supported source, its projection rules, and the sink your workload needs.