Seed + Stream

Copy the starting state without losing what changes during the copy.

SQL Flow pins a source position, plans parallel table slices, records durable completion, and streams the overlap into the same destination pipeline.

Talk through my use case Plan a cutover

Source-aware consistency

Use the strongest snapshot primitive each engine exposes.

PostgreSQL can share an exported snapshot. Other paths combine a pinned log position, replay-safe overlap, durable markers, and idempotent apply.

  • ✓Position pinned before data copy
  • ✓Parallel ranges with bounded concurrency
  • ✓Durable per-table completion markers
  • ✓Resume without restarting completed tables

Prepare the target

Review schema and ingestion before loading.

SQL Flow can inspect the source, describe the expected target, create compatible tables and keys, and journal optional index work.

  • ✓Target schema check and create
  • ✓PostgreSQL binary COPY acceleration
  • ✓Generic heterogeneous loading
  • ✓Confirmed reload and position reset

The payoff

See what the capability changes.

Pinned

CDC handoff

Capture starts at a named position before copying.

Parallel

Bounded workers

Large tables are sliced without unbounded pressure.

Visible

Progress and ETA

Track rows, rate, and completion per table.

Take the next step

Seed once. Keep streaming.

Bring table sizes, keys, source constraints, and target engine. We will select slicing, overlap, and catch-up.

Talk through my use case