Hub for teams replacing Fivetran's managed ELT with self-hosted Airbyte, Meltano or dlt: MAR-based pricing, Hybrid Deployment as Fivetran's own answer to data residency, and the comparison, migration and cost pages under it.
Fivetran against self-hosted Airbyte, whose core is source-available under the Elastic License 2.0 while its connectors are MIT: connector coverage, MAR billing versus infrastructure you run, what the ELv2 license allows for internal use, and the operating work of self-hosting.
Fivetran against MIT-licensed Meltano for teams comfortable with a CLI and Git: Singer taps of uneven quality versus Fivetran's maintained connectors, meltano.yml projects in version control, orchestration with your own scheduler, and who fixes a broken connector.
Fivetran against the Apache-2.0 dlt Python library for engineering teams that prefer pipelines as code: schema inference and incremental loading in dlt, running inside Airflow or serverless functions, and the connector maintenance you take on.
Moving pipelines from Fivetran to self-hosted Airbyte: matching each Fivetran connector to an Airbyte connector, reconciling destination schemas and naming so downstream dbt models keep working, planning initial resyncs, and running both in parallel before cutover.
Moving from Fivetran to Meltano: choosing Singer taps from Meltano Hub for each source, defining extractors and loaders in meltano.yml, handling incremental state, scheduling runs, and checking that warehouse tables match what Fivetran produced.
Fivetran's MAR-based pricing, per-connection base charges and model-run fees against self-hosted Airbyte or Meltano: compute for sync workers, implementation, connector maintenance and on-call time, and the data volumes where self-hosting pays back.
What you can take out of Fivetran before leaving: connection, destination and schema configuration through the REST API or the Terraform provider, what to strip before committing it, and how to use the snapshot as the plan for rebuilding pipelines.
For data engineers who prefer pipelines as Python code: replacing Fivetran connections with dlt sources, recreating schema and incremental loading settings, scheduling runs in an orchestrator you already have, and reconciling row counts before switching off Fivetran.
Fivetran Hybrid Deployment runs pipelines on your infrastructure while Fivetran keeps orchestration, needs an Enterprise or Business Critical plan, and covers only some connectors. This page compares it with Airbyte run entirely in your environment.
Tell us what you're working on, whether it's a deployment, an audit, a security test or a cyber range. You'll speak with an engineer who can help you scope it.