Training non IT users to build reports using DirectQuery over Snowflake
Hi all. I *think* I'm seeing an evolving situation where IT are providing infrastructure, but effectively the project framework they are operating within devolves responsibility over wise use back to the business. "Here's the infrastructure, business users. Fill your boots." This is made potentially more complex/painful by a very strong organisational preference for DirectQuery over import due to an organisational preference to have security handled by the data source.
Chris Webb of Microsoft does a great joint presentation with Craig Collier from Snowflake and Chris Holliday from Visual BI (a partner that specialises in Power BI/Snowflake solutions) here, where they make it clear how important good discipline is in getting a performant report via DQ over Snowflake. In short, touching Snowflake via DQ isn't for noobs. It strikes me using DQ on Snowflake requires discipline, skills likely to fall more into the "Binh the BI Engineer" Power BI persona skillset than the "Ash the Analyst" one. (Will Thompson of Microsoft discusses these personas here.) So even if you provide say aggregation tables that can be mapped to larger tables, you still have to train non IT professionals to know how and why to use them.
At the same time, the Power BI Adoption Roadmap is very clear on the huge investment required in the areas of governance and human capital. I’m worried that IT have completely underestimated how complex Power BI is for even well-seasoned IT professionals to get to grips with in order to produce performant, robust, planned, maintainable reporting, using a wise mix of import, DQ, composite models, and aggs as appropriate to each use case.
I'm possibly going to be the person charged with getting some training in place for these business users, so they can do Self Service BI over this datasource. But my thinking is that no amount of training can turn all business users into career IT professionals…even if all business users had the time and desire.
Any thoughts on this?
Biting my nails. Trying to socialise that Power BI Adoption Roadmap and cultivate a suitable Exec Sponsor to implement it....
- About time that IT gets out of their own way. This whole concept of commodity computing has taken way too long to become accepted reality. Yes, data security and data stewardship are huge topics, and IT can/must continue to play a role there. But business analysis and insight mining should be done by business users together with analysts/data scientists that can translate business speak into ML models.
Two of the most impactful changes in Power BI (in my view) are the introduction of Analyze in Excel and its XMLA endpoint brethren, and the "Personalize Visual" feature. While the former is soothing the fears of the Excel crowd, the latter feature is singlehandedly transforming click-phobic executives into avid information discoverers and Power BI advocates.
- Yes, there is a cost to Power BI for licensing and maintenance, but it is manageable even for larger companies as the vast majority of the infrastructure (apart from the gateways) is in the Azure cloud. As you mention, preventing the data from going walkies is a much bigger challenge, and it requires processes outside of Power BI to be setup as well (up to and including contractual language like standards of business conduct). I don't know if the usage of Power BI or other BI tools increases that challenge or not.