Storage trigger event

[fabricpribeiro]

Dears,

I have a folder in a lakehouse which users use to add some excel files on a daily basis

So far, after adding the files, the users are then triggering manually a fabric data factory pipeline which transforms the excel files into tables in the lakehouse

I would like that the users do not have to manually run the pipeline, but instead, this would be triggered by a storage event trigger from the fabric data factory pipeline  

But some of my team members state that this can become expensive as the trigger will have to keep watching the folder and hence it may cost a lot of money

My undertsanding is a bit diffferent 

Seems to me that its the onelake that will generate an event (created event) which then calls an event grid that will call the pipeline to run it automatically 

This is the definition I saw in AI :

Q1 : Is my undertanding correct? and is this doable for fabric data factory pipelines?  

Q2 : This trigger type (storage) is in public preview or in GA?

Q3 : Can we expect this to be expensive because it will keep watching the folder, or as I expect , its onelake itself that will generate an event and only on file creation? not keeping watching

Thank you so much,

Pedro

[stoic-harsh]

Hey @fabricpribeiro,

Your are already aware of the topic pretty well. But just so it helps others, I will try to answer your questions directly.

Q1 - Is your undertanding correct? and is this doable for fabric data factory pipelines?  

Yes. OneLake events are notifications triggered by changes within the data lake (creation/modification/deletion of files and folders). There is no constant watching happening. OneLake itself fires the event, and the pipeline only wakes up when that event arrives.

The exact flow under the hood is: OneLake --> emits event --> Activator receives it --> triggers the pipeline.

Q2 - This trigger type (storage) is in public preview or in GA?

GA. Azure and Fabric Events (which includes OneLake events) are Generally Available. This was announced in April 2025.

Q3 - Will it be expensive because it keeps watching?

No cost for watching. The architecture is push-based, not polling. So you only pay for pipeline runs when a file actually lands, not for idle watching time. Your teammates' concern doesn't apply here.

And I think, by now, you would have convinced your team😉

[lbendlin]

Reflexes/Eventhouses are very expensive, and not overly reliable. 

Instead, consider running a schedule that only processes files that were added since the last scheduled run.

Or place your Excel files on a SharePoint and use Power Automate to process them.

[fabricpribeiro]

Thanks for your reply

This expects to work on 3 projects. Each project adds 20 files por month, each file with 2K rows, not more then that

Do you still consider it expensive which such a low number of rows? or you think its ok?

My capacity is F16

[lbendlin]

The triggers aren't expensive. Keeping the Eventhouse running is.

[fabricpribeiro]

Here are the monitoring costs

They seem very, very low? and its running on a F2 capacity