Docs/Streams/Concepts

Convert Mode

Convert mode is the bounded data-transfer mode in DBConvert Streams. It reads the selected source scope directly, writes the result to a target, and finishes when the transfer is complete.

When to use Convert mode

Use Convert when the job should finish after transferring the selected data — an initial load, a one-time sync, or a filtered export. For a side-by-side comparison with CDC, see CDC vs Convert.

How it works

DBConvert Streams reads the selected tables or a custom SQL query from the source, applies the configured stream settings (structure handling, data policies), and writes to the target. The stream stops when the scope is complete or a configured limit is reached.

Convert supports a broader set of source and target combinations than CDC. Sources include databases, local files (CSV, JSON, Parquet), and S3-compatible storage. Current targets include MySQL, PostgreSQL, files, and S3. Snowflake target support is documented separately and is coming soon. Convert also supports multi-source streams — combining data from several connections in a single transfer. Confirm your combination in the Capability Matrix.

Source configuration by engine

Source engineConvert configuration guide
MySQL-compatibleMySQL Convert Mode Guide
PostgreSQL-compatiblePostgreSQL Convert Mode Guide

What Convert does not do

  • It does not capture ongoing changes after the transfer finishes — use CDC mode for that.
  • It does not keep running after the selected scope is transferred.
  • It does not propagate DDL changes made on the source after the stream starts. However, it does create tables, structures, and indexes on the target if they are missing — controlled by the structure options in the stream config.

Custom SQL queries

Convert mode supports query-driven extraction — filter rows, select a subset of columns, or shape the exported dataset with a custom SELECT. See Custom SQL Queries for syntax and examples.

Typical rollout with CDC

A common production pattern:

  1. Run Convert for the initial load to bring the target up to date.
  2. Create a CDC stream on the same source to capture ongoing changes.

This only applies when the source–target combination supports CDC.