cancel
Showing results for 
Search instead for 
Did you mean: 
Reply
Highlighted
Cymbolz Regular Visitor
Regular Visitor

Dataflow vs Dataset refresh

Haven't found any documentation on how data refresh works with respect to a dataflow and then a dataset sourced from that dataflow.

 

So looking for feedback, based on what I've discovered:

 

  • Both a dataflow and dataset need data to be refreshed
  • So I assume the dataflow is much like a data storage component on it's own that manages the updating from the data source, wherever that may be
  • And the dataset will refresh data from the dataflow 'storage'
  • Thus a logical refresh sequence (such as setting a scheduled refresh) would see the dataflow update first then the dataset aftewards (maybe 30 mins later as I suspect doing both at the same time may not yield the right results)

I've come to this conclusion after seeing the behaviour of having one or the other set for scheduled refresh.

 

I'm also seeing inconsistency in the workspace contents view where it shows last and next refresh times.  

 

For this dataflow, I've toggled off the scheduled refresh but it still shows a Next Refresh time (I would expect not to see any time stamp):

Capture 1.PNG

 

For these datasets, they have both had a refresh more recently than indicated here

 

 Capture 2.PNG

 

Here's the first one:

Capture 3.PNG

And the second:

Capture 4.PNG 

 

A bug?

 

2 ACCEPTED SOLUTIONS

Accepted Solutions
otravers Member
Member

Re: Dataflow vs Dataset refresh

>So I assume the dataflow is much like a data storage component on its own that manages the updating from the data source, wherever that may be

 

That's correct, technically PBI's Dataflow uses Azure Data Lake Gen2 for storage.

 

One use case I plan to use this dual refresh structure for, is to handle sources (e.g. static files) that don't need to be refreshed in Dataflows where they'll be imported but not under scheduled refresh. I've found PBI's scheduled refreshes to fail easily, so cutting down the service's scheduled refreshes to sources that actually need to be refreshed should lower incidents (e.g. web API timeouts, credential issues etc.).

View solution in original post

Assaf Frequent Visitor
Frequent Visitor

Re: Dataflow vs Dataset refresh

Hi,

 

After internal checking, it seems that the next refresh time update (without browser refresh) was fixed and is should be available in the following updates.

 

I will keep monitor it.

 

Thanks,

Assaf

View solution in original post

16 REPLIES 16
otravers Member
Member

Re: Dataflow vs Dataset refresh

>So I assume the dataflow is much like a data storage component on its own that manages the updating from the data source, wherever that may be

 

That's correct, technically PBI's Dataflow uses Azure Data Lake Gen2 for storage.

 

One use case I plan to use this dual refresh structure for, is to handle sources (e.g. static files) that don't need to be refreshed in Dataflows where they'll be imported but not under scheduled refresh. I've found PBI's scheduled refreshes to fail easily, so cutting down the service's scheduled refreshes to sources that actually need to be refreshed should lower incidents (e.g. web API timeouts, credential issues etc.).

View solution in original post

Assaf Frequent Visitor
Frequent Visitor

Re: Dataflow vs Dataset refresh

Hi!

 

Regarding the inconsistency in the next refresh time of the dataflow, after a browser refresh, do you still see the next refresh time?

 

Thanks,

Assaf

Cymbolz Regular Visitor
Regular Visitor

Re: Dataflow vs Dataset refresh


@Assaf wrote:

 

Regarding the inconsistency in the next refresh time of the dataflow, after a browser refresh, do you still see the next refresh time?

 


I do see the refresh time after a browser refresh.  Also with the page remaining on screen, the refresh was scheduled to run and the date/time stamps updated without me having to refresh.

 

The issue I was experiencing included navigating away from that pgae, then returning to the page (so not refreshing the browser), so I'd have thought that would result in the updated time stamps too...I'll keep an eye on it.

Assaf Frequent Visitor
Frequent Visitor

Re: Dataflow vs Dataset refresh

Hi,

 

After internal checking, it seems that the next refresh time update (without browser refresh) was fixed and is should be available in the following updates.

 

I will keep monitor it.

 

Thanks,

Assaf

View solution in original post

Cymbolz Regular Visitor
Regular Visitor

Re: Dataflow vs Dataset refresh

I've noticed the date & time stamps are now updating, even without any refresh or navigating away.

Assaf Frequent Visitor
Frequent Visitor

Re: Dataflow vs Dataset refresh

Good to hear. Thanks!

Lasselakan Regular Visitor
Regular Visitor

Re: Dataflow vs Dataset refresh

Do I understand it correctly that you found out that both the Dataflow and the Dataset needs to be updated (either manually or scheduled). That's a quite unfortunate dependency... I expected the Dataflow to need to be scheduled, but the the Dataset to be "DirectQuery".

BIbigbox Frequent Visitor
Frequent Visitor

Re: Dataflow vs Dataset refresh

I do love dataflows and what they allow us to do (preparing entities only once and using them everywhere is a dream come true).

However, that's really disappointing... I will now suffer from more delay: the dataflow will have to refresh first (let's say at 9:00), and half an hour later the dataset (9:30). I wish datasets would automatically refresh when the dataflow does...

skcary Regular Visitor
Regular Visitor

Re: Dataflow vs Dataset refresh

Yes I don't understand this either. Hopefully someone can elaborate. Even if my report only connects to a single dataflow, it still creates a "dataset" when I publish it and thus I have to manage two refresh schedules (one for dataflow and one for dataset). I can see some use cases but I would think most of the time the preferred functionality is for a dataflow refresh to automatically kick off the associated dataset refresh.

Helpful resources

Announcements
Can You Solve These Challenge

Challenge: Can You Solve These?

Find out how to participate in the first Power BI 'Can You Solve These?' challenge.

Community News & Announcements

Community News & Announcements

Get your latest community news and announcements.

Virtual Launch Event

Microsoft Business Applications October Virtual Launch Event

Join us for an in-depth look at the new innovations across Dynamics 365 and the Microsoft Power Platform.

Community Kudopalooza

Win Power BI Swag with Community Kudopalooza!

Each week, complete activities and be qualified in the drawing for cool Power BI Swag.

Users Online
Currently online: 137 members 1,504 guests
Please welcome our newest community members: