My wife rolls her eyes every time I pull out Microsoft’s Power BI to analyze our personal bank statement data. Perhaps she doesn’t want me to know how much she spends each month on Starbucks Grande Chai Lattes… no whip. The reality is, I utilize Power BI whenever possible. In fact, these days I find it hard to engage in a business intelligence strategy discussion without Power BI being part of the conversation.
In recent weeks, I've attended many Power BI events, presentations, local user groups, and even our own Power BI Meetup. As I prepare to speak on the topic of Power BI at upcoming PASS SQL Saturday’s this fall, I’ve discovered an alarming trend with the way organizations are preparing for scalability and governance of a Power BI solution. A calculated approach to the initial deployment is critical to ensure success down the road. It is an alarming trend which, if not immediately affecting your business, soon could be. Before dissecting my concerns, I’d like to lend some context to this through a summary of some of the common usages of Power BI.
Power BI really is multifaceted product, but I perceive its primary function today is as a reporting tool first with self-service analytics for those who want it. 95% of organizations believe end-user adoption of self-service analytics is the future. It translates to many end-users utilizing this tool, many of which have a mountain of knowledge in their industry segment. The end-users themselves can determine exactly what he or she wants to see which in turn goes a long way to removing the commonly identifiable bottleneck of getting reports from IT. Microsoft has enabled these users to use a tool that is genius in its simplicity and increases speed to insight and better decision making.
With that said, I return to my original concern over deploying a Power BI’s solution without properly considering scalability and governance. This trend that has begun in some larger organizations, where many of the models being created are growing organically for solutions that are quickly becoming unsuitable and unstable as organizations scale. While Power BI resources can be scarce, Power BI adoption and end-user acceptance happened quite rapidly, which can be both a blessing and a curse.
Here are 5 Obstacles to plan for with a Power BI deployment:
- Organic Power BI growth usually stems out with little or no plan on scalable infrastructure
- Measure based table additions inflate the number of measures, both simple and complex
- Additional measures get added over time and replicate in multiple tables
- Data volumes increase over time and require a change to the model and calculations
- Security & governance become dependent on sound architecture principals
When built with scalability in mind, these concerns would barely cause Power BI to blink. However many initial offerings are not designed for scale and when complexity goes beyond the initial concept stage, this usually stays as the production entity. When this occurs, you are guaranteed to start seeing (1) refresh times get longer and (2) report rendering performance decline. Self-service analytics decreases as users rely on the data model to be fine-tuned for optimal performance. Is this a problem with Power BI the tool, or is it the underlying architecture/model that need to be addressed? Without proper training, the end-user won’t be able to tell the difference. As an organization moves forward with a culture of self-service adoption, performance issues could instantly kill all the momentum companies have works months to build.
These are just a few of the symptoms I’ve identified through my own experiences. Now, to be clear, those who have thought through these items carefully prior to handing data or models to the business, should not be having any issues because they were built properly and with scalability in mind. These issues can be fixed but as you would expect the longer you leave it and the more you patch the problem, the more complex and more expensive it will become to resolve. Vetting a trusted Microsoft Gold Partner & strategic advisor to help with a Power BI solution from the very beginning can ensure the greatest long term savings that will pay itself back over and over again.
George Bryant is a Senior BI Architect at Beyond Impact, a Microsoft Gold Partner based in Minneapolis. George has over a decade of experience in the Business Intelligence industry and is an active member of the BI community. George specializes in BI architecture and end to end development in order to deliver advanced analytics and metrics to multiple organizations across the country.