Azure: Is Azure Data Factory (ADF) Expensive to Use?

I’ve heard several people complaining that Azure Data Factory (ADF) is expensive to use, and yet I’ve seen many others talking about it as a really low cost service.

So which is true?

Our experience

We’re very careful about which features of ADF that we use as there are a few that can increase the cost substantially. And without those, it becomes a very low cost service.

None of my current clients has any issue with ADF costs. The people that I hear complaining about the cost are using one or both of:

SSIS Integration Runtime
Dataflows

SSIS Integration Runtime

When moving to cloud-based services, many people want to just move their SQL Server Integration Services (SSIS) packages and not need to modify them.

I understand the sentiment. But it can potentially greatly increase your ongoing costs.

It’s quite remarkable that the team have created a way to run existing SSIS packages in ADF. But that doesn’t mean you should use it. Even if you do move them, there are usually things you need to do to get those packages working. It’s not just a lift and shift. But that’s not the main issue.

What you are doing by moving SSIS packages into ADF is trading off a short-term cost for an ongoing cost that’s potentially much larger. The SSIS integration runtime is expensive to keep running.

While in some cases, it’s been hard work to migrate existing SSIS packages to native pipelines in ADF, it’s my preference almost 100% of the time. And what you end up with is usually superior as well.

Dataflows

Dataflows in ADF are convenient and easy to use. They let you work with an equivalent of Power Query right within ADF.

But again, you pay a long-term price for the short-term development price. If you’ve used them, you’ll also notice that they’re much slower to “get running” as well.

We avoid using these.

Note: I’m not talking about the Gen 2 dataflows in Fabric here, but the dataflows that are in ADF.