Describe the pipeline.
Choose source, projection, target, runtime role, and secret references in validated configuration.
Capture changes from the systems you already run, maintain joined search projections, and deliver live events over GraphQL or native WebSockets.
Self-hosted data plane · Zero-downtime backfill and cutover · No application dual-writes
Capture · join · denormalize · checkpoint
12:42:08.114 cdc event received relation=orders 12:42:08.117 projection composed children=4 12:42:08.120 sink acknowledged version=94821 12:42:08.120 checkpoint advanced lag=0
Your database already keeps a perfect record of every insert, update, and delete — its write-ahead log. Change data capture (CDC) reads that log directly, turning each committed change into an event the moment it happens. No polling, no “also update the search index” code scattered through your app, no drift between two writes that were never in the same transaction. The same mechanism doubles as zero-downtime migration: bootstrap the existing data in bulk, tail changes until the new store is caught up, then cut over.
Each capability has a focused path through its architecture, operating model, and deployment boundary.
Keep denormalized search documents synchronized from native database changes.
Explore CDC 02 / DELIVERYDeliver durable broker events through GraphQL and native WebSockets.
Explore realtime 03 / MIGRATEBootstrap existing data at bulk speed, then tail live changes until you cut over — one code path, nothing to drift.
See how 04 / OPERATEUse our hosted control plane to administer customer-hosted agents across organizations and environments.
Explore FleetChoose source, projection, target, runtime role, and secret references in validated configuration.
Start a native binary, Docker container, or Helm release in your environment.
Apply revisioned desired state and administer every environment from the CLI.
Run VentStream beside your data, validate the result against your own workload, and add Fleet when you need centralized operations.