On premise IT managed data model business develop report

[luke_jah]

Hi

I'm relatively new to Power BI. We are piloting the following:

- Power BI On premise licensed through SQL Server Enterprise w SA.

- Data source on premise from data warehouse

Looking for the best practice way to deploy:

- IT managed data model

- Business create reports

In my readings, the advice is to use workspaces and content packs. However I belive those are only available in the Power BI service (ie cloud option). What is the equivalent for the on premise or workarounds?

Thanks

[Anonymous]

We are doing something similar here.

We are using published Tabular models and Live Connections on all our reporting that is managed. We also have "published" views onto some of our operation SQL data sources that we again reccomend are used through Direct Query.

We publish a set of "template" PBIX files that have the relvant connections in as starting points for reporting. These also have our Branding in them.

The freeform mashups are an entirely different bag of rabid badgers! Oh and users will try and import your entire SQL DB/TAB model so they can mash it up with some Excel. Provide them with cutdown datasets for this kind of import. It gets really really boring explaining why their data import is slow (12 years worth of data...)

Self Service reporting is just so much fun! when I say fun I mean in the walking a tight rope over a tiger pit without a safety net (with the rope on fire) kind of way.

[luke_jah]

stpnet thanks for the feedback and excelent visuals on the tightrope.

Great idea with the template PBIX with the connections details.

Keen to know:

- how you manage the template (is it one template per view?)

- how you manage the approved views

- how you manage the performance as it is direct query

[Anonymous · Accepted solution]

We have one template per model/dataset. We do have a couple of different tempaltes for our more complex data sets which are functionally oriented. So you pick the right template for the functional area you are looking at, or the aspect of a dataset you are looking at.

We monitor the performance of the SSAS cubes with a mix of perfmon metrics that we gather off the servers and SSAS Monitor and SSAS Events Analyzer (both on sqlbi.com) the perfmon metrics we lift are based on the MS Tabular whitepaper and some blog articles. Mashing them together is pretty untidy. Oddly enough we do it in PowerBI 🙂

The approved views are a conventional BI solution (DB backend to Tabular Model) that runs through the usual specification, version control, change management etc. So way slower than the users like but better managed and more rigourously tested/designed. The eternal trade off... Cheap, Fast, Right, pick any two!

S

[luke_jah]

Thanks again for the ideas. We'll check these out and see how well this works.