Warehouse Performance

#1

We have several process-heavy data cubes that we warehouse. This works well in general, except when the warehouse job is running, no data is returned on the dashboards. Is there any way to ensure that the warehouse build(s) don’t knock out the entire dashboard while they are running? It would be nice to set it to show cached/old data until the data is refreshed, but I’ve read a few of the support articles, and can’t figure it out.
I believe incremental warehousing won’t work in this case because we need to incorporate data changes as well as new data.
Long term, I would like make the data more streamlined coming from the source to improve performance, however that is not an option at the moment and I am relying heavily on joining and calculated elements in the cube.

Thank you

1 Like
#2

I would love to have this too. I believe that this used to be the case until it got changed in v6.

1 Like
#3

Agree with both. V6 introduced something weird with warehousing. luckily it did not effect me much, well my users but it cause my development work a lot of pain

1 Like
#4

The problem occurs only when the cube is warehoused the first time. After that, during warehousing time, the data will be returned from the old warehouse. Same thing for in-memory cubes.

We do not retrieve data from the source during first time build simply because, for many users, such an action will lead to massive CPU/memory overloads.

#5

That has not been our experience. We have hourly scheduled warehouse jobs and during that time, there is a 3-5 minute period where our dashboard solution just errors out. Is this not how it should work? Is this something I should take up with support?

#6

Yes, I think it warrants investigation.

#7

@adrian.dobrin Not my experience either.
Should I also take it up with support? Or just wait to see if @molly.loughran gets an answer that I can use?

#8

You can open a ticket as well. Once built, a warehouse should always allow queries, irrespective of the current build status.