Contents

Data Engineering › Ingestion

Managed Ingestion Connectors

Tools like Fivetran or Airbyte that sync SaaS and database sources for you.

Also known as: Fivetran, Airbyte, ELT connectors, managed connectors

Managed ingestion connectors are prebuilt integrations that copy data from a source (Salesforce, Stripe, Postgres, Google Ads, and hundreds more) into your warehouse. Tools like Fivetran and Airbyte provide them, and you configure a source, a destination and a schedule instead of writing code.

Source: Stripe  ──connector (hourly)──▶  warehouse.raw.stripe_charges, ...

The tool deals with the unglamorous parts: authentication, paging through the API, rate limits, retries, incremental syncs, and often schema changes. Data usually lands “raw”, and you transform it afterwards (the ELT pattern; see ETL vs ELT).

When it’s a good fit

  • A common SaaS source that thousands of other teams also use, so the connector is well-tested.
  • A small team that would rather spend time on modelling than on maintaining API scripts.

Trade-offs to think about

  • Cost. Many are priced by data volume or rows changed. A chatty table can get expensive, so watch usage.
  • Control. You can’t always change how data is extracted, which fields are included, or how deletes are represented.
  • Gaps. A connector may not cover an endpoint or custom field you need, and then you build that part yourself.
  • Schema changes. Source changes can appear as new or altered columns. Plan for schema drift.
  • Compare the alternatives. Hand-written pipelines give full control but you maintain them. See build vs buy.

Habits

  • Keep the raw copy untouched and do cleanup in separate models.
  • Monitor sync failures and freshness; “no data arrived” can be silent.
  • Understand whether a connector does full or incremental loads, and how it detects deletes.
  • Check the vendor’s current docs, since behaviour and pricing change.