Azure Data Factory is a managed, serverless ETL tool with a drag & drop UI for use in the Azure cloud. It is a good product, but lacking in maturity in some areas - mainly surrounding the UI itself.
- Drag & drop functionality makes simple tasks easy.
- Managed service — no infrastructure to manage in-house for running ETL processes, or orchestration of them.
- Integrates well with other Azure services, external APIs, and on-premises data stores.
- DataFlow within Data Factory is essentially "managed DataBricks", which scales well for heavy data processing tasks, again configurable through the UI.
- Exports to an ARM template for automation.
- There are still several bugs in the UI, which can cause frustration while debugging.
- Working from the UI is very "manual" — but developing the jobs-as-code is too slow, as this means changing the ARM template, redeploying and running.
- Lack of online resources from the wider community, so less to learn from other's experiences. This should change as adoption increases.
- Price can be higher than other solutions, when coupled with "DataFlow" — but DataFlow is where the bulk of transformations happen. DataFlow is better suited to heavy processing tasks, rather than basic/simple transformations — consider other options here.
It's worth remembering that Data Factory is under active development, and it has great potential as a product when these "teething bugs" are ironed out. These observations were last updated: 14th February, 2020