Datset now too big to be able ot apply query changes in desktop so cannot publsih

[masplin]

My model has been runing for over a year and during that time one table has got pretty big.  i was making some adjustments and it was taking ages to apply the query so i killed desktop in task manager (as cancelling seemed to take forever as well). When I reopened the file all the data is gone from the data tables. I refreshed each table one at a time, but the main table will just not refresh (after 8 hours of churning).

i tired saving, clicking apply later, and then publishing. I then refreshed on the service, but this table data is missing whne it completes. Clearly service can only refresh the applied queries. 

I'm now completely stuck as cannot complete the apply step in desktop.  The machine I have to use to conenct to the source is a virtual machine that doens't have sufficient RAM.  i can't connect to the source form any other more powerful machine.

Any idea how i can move forward?  I can cut the date range down to this year in desktop, but this seems to first try and load all the entries so no quicker.  Anyway applying that loses all my YoY calculations.

Going on holidya in 2 days and whole company reporting is knackered!!!  Any help appreciated

Mike

[v-yulgu-msft]

Hi @masplin,

As you know, this problem is caused by the too large dataset. Also, if the size of your pbix file is over 1 GB, it would failed to publish it to service. Per my opinion, the only way is to reduce the data amount loaded to desktop. Please check whether this blog is helpful to your scenario.

By the way, what is your data source?

Regards,

Yuliana Gu

[masplin]

Well oddly it seems the problem is actually power bi desktop. I spent hours trying to get the one table to refresh with no luck.  Only 1mio rows and workbook is only 75Mb.  However i use power Update to update my pbix at night and that seems to have no trouble at all!!!  Don't understand why it works and manual refresh doesn't.

If it hadnt worked, my next idea was to disable al lthe relationships on that table as reduces the mount of computation to do the refresh.  Once the table is applied oyu can then tunr the relationships back on.  Better to try that than cut down the dataset.