Monthly Archives: August 2016

Is Power BI ready for the enterprise?

Power BI Ready For The EnterpriseWith the latest Power BI releases in mid-2016 Microsoft finally plugs the gap between on-premise Business Intelligence solutions and the cloud. On-premise enterprise BI solutions can now be re-used seamlessly with low level of development and infrastructure changes. If you would have asked Microsoft, they would probably have told you that Power BI was enterprise ready years ago. Even though the product might have been visually appealing and worked well for self-service, small businesses to SME’s the product lacked important features for enterprise BI. Scalability and stability have also been an issue. Until now. Last year Microsoft have made quite a commitment to Power BI and some important features have been added which makes Power BI more mature and maybe a game changer for enterprises to start moving Business Intelligence into the cloud. Among some of the exciting features that have been released last year are:

  • Live connection from Power BI to existing on-premise multidimensional cubes using On-Premise Data Gateway
    – No need to rebuild cubes to tabular cubes anymore. This also removes any scalability issues of having data models stored in the cloud (limits on data size and data refresh timeout). Re-use your existing multidimensional cubes and infrastructure.
  • Analyze in Excel
    – Users can now connect Excel directly to the Power BI service for data either stored directly in the cloud or your on-premise multidimensional or tabular cubes. Functionality for Power BI Excel connectivity to multidimensinoal cubes where added last month. This is a very important step since now we are able to benefit of interactive visuals and collaborative features of Power BI and also at the same time keep existing functionality of Excel for end users. 
  • Reporting Services Integration
    – Re-use your existing Reporting Services reports in Power BI by pinning them to your dashboards. Currently you will only be able to access them when you are within your internal network.
  • Integration with Azure Machine Learning, Azure Data Factory and statistical analytics using Microsoft R which will open more business opportunties through your data.
  • Integration with Big Data/Hadoop/HDInsight and all kinds of other sources
  • Cloud based BI solution for a single point of entry to all our enterprise data at an affordable price
  • Improved stability and reliability
  • Good drill-down functionality within the same dashboard. (This is something that PowerView and Power BI lacked previously)
  • Mobile BI and custmoization of reports for mobile view (This part still leaves some room for improvements. For example Android tablets are not supported yet, however Android phones works well)
  • Role and row level security. Possible though using live connection to on-premise analysis services and re-use existing security models or directly within power BI itself (the latter is still i preview).

With the features above we have the ability to create BI solutions that cover the whole span of analytics from traditional reports, interactive dashboards, mobile BI and self-service BI both for advanced (using Excel/R) and mainstream end users (using easy drag and drop functionality in Power BI).

I find that many enterprise organizations that use Microsoft as a BI platform tend to rely heavily on Excel, Reporting Services and SSAS multidimensional cubes. Hence I find it very strange that full integration and compatibility with those on premise solutions not have been incorporated into Power BI from the first release. However now it is here so if you belong to the organizations who use Microsoft BI and are considering moving into the cloud Power BI can be the perfect match.

Over the years Microsoft BI portfolio have been going in all directions with multiple solutions that a lot of time overlap each other making it hard to know which tool to use where and for what. Hopefully we can now start to see the end of it. Just to mention a few BI visualization tools from Microsoft:

  • Performance Point Server
  • Reporting Services (standalone)
  • Reporting Services (through SharePoint)
  • Performance Point Services (through SharePoint)
  • Excel Services (through SharePoint)
  • PowerPivot
  • PowerView (though Excel)
  • PowerView (through SharePoint)
  • PowerMap (through Excel)
  • Plain Excel
  • DataZen (which now is part of reporting services)

*By SharePoint I refer to SharePoint Server on-premises

Microsoft have for some time had a strong backend for BI solutions using Microsoft SQL server and Analysis Services. However, before Power BI Microsoft have been behind competitors when it comes to visualize data in an easy and good looking way. Either organizations have used another vendor or relied on multiple solutions from Microsoft to achieve same features such as combining SharePoint, Excel Services, Power View and Performance Point services. The first usually require additional licensing costs and the latter also add licensing costs but also costs for maintenance and overhead since having a flora of multiple solutions increase complexity.

I think with Power BI getting more mature it is a good opportunity to see how existing solutions can be consolidated to reduce cost and make maintenance and scalability easier. Combining Power BI, Reporting Services and Excel would cover most organisations visiualization needs and could be used to repalce various on-premise solutions such as SharePoint BI and third party solutions. If you already are on Office 365 you might consider to scrap the entire on-premise SharePoint farms all together. Far to often I believe SharePoint farms are installed with limited knowledge and never used to their full potential. However, the product is rather complex and tend to be maintenance heavy and error prone if not taken care of properly. Hence Power BI may be an appealing option since we do not need to care for infrastructure, configurations and security patches etc. There might be cases where data is required to be kept on premise for legal reasons or complex customized solutions so that we cannot move away from on-premise SharePoint. However with Power BI we are now able to let data stay on-premise and do the data visualization, gather insights and collaborate and share them in the cloud.

When I first got in contact with Microsoft Power BI in 2013 the product felt as an unfinished implementation of PowerPivot and Power View in Office 365 rebranded as “Power BI”. Even though the solution provided a nice facelift to Microsoft BI portfolio it lacked a lot of functionality that would be required for the typical enterprise and also had a lot of stability and scalability issues and lacked important features for end-users such as:

  • Not being able to re-use existing multidimensional cubes. Only tabular cubes where supported and they had to be created within Excel and PowerPivot and then uploaded to the cloud. Later functionality to access on-premise tabular cubes where added but multidimensional cubes where left out. I find this very strange since if Microsoft would have liked a quick adoption of their product I believe they should have had supported multidimensional cubes first since that is what most of their existing clients would already be using.
  • Not being able to re-use existing reporting services reports.
  • Limited visual customization due to the use of Power View (for example you could not even properly customize the colors of a chart, you could only use pre-defined themes)
  • When keeping PowerPivot data models in the cloud there was initially a limit of 2gb. Also to schedule them for refresh had a 30-minute timeout. Hence analytics of larger datasets where completely out of the picture. Going with this approach usually also resulted in that existing multidimensional cube solutions had to be rebuild as tabular models in PowerPivot.
  • A lot of bugs and random errors. (I have a 22-page word document with screenshots of different errors which I send to Microsoft)
  • No native apps for Android or IOS only windows
  • Based on Silverlight, then an effort to make it work in HTML5 was done however it lacked functionality compared to Silverlight function
  • The HTML5 version did not work properly on iPad2 (It crashed when you did any kind of zooming with your fingers in Safari)
  • No proper drill-down functionality (All visual objects could only show data from the selected level and drill-down had to be achieved by having multiple reports without connections to each other)
  • No good way to restrict data for certain roles/users other than creating multiple reports/data sources

Current version of Power BI has solved the issues above and feels as a more mature product. Microsoft already had a strong backend for BI solutions and with Power BI they have now bridged the gap between the two. The service comes with 99.9% uptime, financially backed service level agreement (SLA) that will be enough for most organizations. Enterprise organizations are now able transition into the service by re-using their on-premise data warehouse (DW) and analysis services solutions at an affordable price making Power BI a competitive choice in the BI market.