PBI Desktop refresh differs from PBI Service refresh.

[Vallirajap]

Hii,

In recent times, I have noted a dataset refresh timing difference between the desktop and service.

What is the reason when a dataset gets refreshed too late in the PBI service?

Sometimes it fails after running for 2 hours.

If anyone knows tell me the answer,
Thank you

[ibarrau]

Hi. To keep it simple let's say there are 2 or 3 factors that would affect a refresh.

1- The source performance. A database might be faster than reading a file. But it can be a complex view that take times to read it. The time that takes the source to response is key. There are APIs that run fast and other that take a lot. Some sources can be tested. If you have a query for reading a database you can test it in a DBMS like DBeaver.

2- The transformations. The more complex your power query cleaning and wrangling is, the more time it can get. Optimizing your power query transformations steps can help reduce the timings.

3- Bandwidth (optional): if you have some sources that depend on a Gateway, then you might consider a third variable that is the bandwidth to move data. Some countries still have internet providors with small bandwidths for large datasets.

You can take a look at those three variables. The refresh fails because Power Bi Service has a 2 hours window at a Shared Capacity Workspace to refresh a semantic model.

I hope that helps,