Power BI is a hot topic within the Microsoft Business Intelligence community. Since it was announced last July at the Worldwide Partner Conference in Houston, it’s pretty well been the only thing that has gotten discussed with regard to Microsoft BI. There are good reasons for this, the addition of a mobile BI story, its ease of use, and it’s compelling new features (Power Q&A!) make it the shiny new toy. I’ve certainly been paying a great deal of attention to it, but what about the more traditional products, in particular Power Pivot for SharePoint?
One of the big questions around Power BI is whether or not there will be an on premises version of Power BI. Whenever asked, Microsoft responds with “Power BI will initially be available through Office 365”. This answer causes concern to those with requirements that can not or may not be met by a cloud based solution. Many worry that Microsoft’s move into devices and services are leaving on premises installations behind. I’ve been of the opinion that this is evidence of a “cloud first” release strategy, as opposed to a “cloud only” release strategy. Recent statements by Microsoft officials would tend to confirm this, but the question should be asked, does it even make sense to bring Power BI on premises?
A little explanation is in order. The collection of tools that is Power BI is centred around the x-Velocity data model that is part of Excel 2013, and available through Power Pivot in Excel 2010. All of the client based design tools can be used with Excel without the need for a Power BI license. With Power Pivot for SharePoint, it has been possible to interact with these data models through a browser for several years already, and with the 2013 updates to Office 365, it is even possible to interact with these models in the cloud. What has been missing from the Office 365 BI story has been an automatic way to keep on premises data refreshed, and the ability to work with large models.
While these two capabilities have been available on-prem for years through Power Pivot for SharePoint, they are only coming available to Office 365 now with Power BI. It doesn’t really make sense to bring these capabilities on-prem when they already exist. However, complicating this picture is the host of new capabilities that are being introduced by Power BI. In many ways, it’s a “leapfrog” product, filling in gaps in some areas, while moving forward in others. A comparison of the two products can be seen below.
Let’s walk through these features. Obviously both products work with the embedded x-Velocity data models. Power Pivot for SharePoint from SQL Server 2012 SP1 can render Power View in Excel, as can Power BI. Power View has some interesting variations however. Through the Power View that is available via Power Pivot gallery, live Power View reports can be exported to PowerPoint decks. This feature is not available through Excel Power View, or through Power BI. On the flip side, on-prem Power View reports (both types) use Silverlight for rendering, whereas Power BI will allow both Silverlight and HTML 5 renderings (confused yet?).
Both Power Pivot for SharePoint and Power BI are powered by an Analysis Services engine. The Power Pivot gallery available on-prem provides for the ability to connect to that engine through Excel with an Analysis Services connection. This makes the embedded model created in one workbook available to Excel clients as what appears to be a data cube. This is not available through Power BI, although the OData publishing features fills that gap somewhat.
The ability to refresh the data in the embedded model is critical and is to my mind, the most important feature in Power BI. However, at best, this brings it to parity with Power Pivot for SharePoint. For the moment it supports only SQL Server on-prem data sources where Power Pivot for SharePoint supports all Power Pivot data sources for refresh. As of this writing (December 2013) neither product supports the refresh of Power Query data sources, but this has been promised for Power BI “soon”. No announcement has yet been made as to the refresh of Power Query data sources on-prem.
The default maximum file size for SharePoint 2013 is 200 MB, and the default maximum workbook size for Excel Services is 10 MB. These values can be changed on prem, making the maximum possible size for a data model equal to the maximum possible file size in SharePoint – 2 GB. This equates to the maximum file size in Office 365 as well, but that 10 MB Excel Services limit can’t be changed in Office 365. Power BI supports model sizes up to 250 MB by removing the model portion from the workbook, and housing it in an Analysis Services instance, allowing the workbook to remain within the 10 MB limit. It’s wonderful to be able to move beyond the 10 MB limit that we’ve had, but it’s not without its limits.
Both products have a thumbnail gallery, but the one available through Power BI sites is arguably more sophisticated, and it doesn’t rely on Silverlight for rendering. The rest of the feature set outlined above is all in Power BI’s favour. Power Pivot SharePoint will optimize workbooks (move the data model into Analysis Services) on first interaction, but Power BI can do that ahead of time, minimizing user inconvenience. The rest of the feature set, OData feeds for on-prem data, Power Q&A, and mobile clients are only available with Power BI.
The fact that Power BI for SharePoint on premises has not been announced isn’t as disappointing as it may seem. Parts of it, the Data Management Gateway in particular, aren’t even necessary in an on-prem scenario. This is pure speculation on my part, but if I had to bet, I would expect to see the relevant features from Power BI (Power Q&A, OData publishing) put into Power Pivot for SharePoint in a future release of that product. It also wouldn’t surprise me to see it renamed to Power BI for SharePoint. As to when this could happen I have absolutely no idea, but we should keep in mind that this is a product that ships with SQL Server, not with SharePoint, and I haven’t heard of anything like this in the previews of SQL Server 2014.