Key Features in the New Power BI Service Announced

As posted previously, Power BI is getting a make over from its first version launched early last year.  Microsoft has just announced some new details on the revamped business intelligence service.

Power BI is going Freemium for Consumers, Discounted for Businesses

Microsoft has announced that there will be a free version of Power BI.  In addition, there will be an enterprise version called Power BI Pro that will replace the existing subscription pricing available in Office 365 today, but at a significantly reduced price of $9.99 per month.

Power BI is being Disconnected from Excel

Up until now, the key design surface for Power BI has been Excel 2013 with its Power Query, Power Map and Power View add-ons. 

In the new version of Power BI, Excel is no longer a requirement for building dashboards – you can build them through a browser.  You can still design dashboards using Excel and upload them to Power BI but this is no longer a requirement.

Instead, the new Power BI Designer will allow you to build new dashboards.  From the current model, the Power BI Designer is still an application that needs to be installed and not a web application and is only currently available on Windows.

Power BI is being Disconnected from Office 365 and SharePoint

The history of Power BI is rooted in Excel Services and SharePoint.  One of the key reasons to use SharePoint as a portal was because it was the vehicle to run excel services.  Excel services allowed you to render excel reports, dashboards and charts without the need for an Office desktop client.  This was one of the key features of SharePoint – see the old SharePoint 2010 “wheel” and you’ll see “insights” (e.g. Excel Services and BI) is a key part of the feature set.

Well in a similar strategy that has already been seen with One Drive and Office Web Apps, Power BI will be no longer tied to SharePoint and exist on its own as a set of apps, services, APIs and data.  You’ll still be able to presumably add the Power BI App to your SharePoint sites, but you’ll just as easily be able to go directly to the Power BI App and consume reports directly.

Power BI will have New Visualizations

As part of the new Power BI, dashboard designers will have access to new visualizations including:

Combo Charts

Gauges

Tree Maps

Filled Maps

Funnel Charts

Power BI is still Cached Data and Refreshes are Limited

The core of Power BI is Power Pivot and its in memory cached data repository.  This architecture still holds true in the new version which means that if you’re looking for real time streaming, direct access to your data sources, etc. then you’re out of luck.  In addition, while Power BI’s caching feature is highly useful for data that is cacheable (e.g. static reports, weekly dashboards, distributed data extracts, etc.) it still has some weaknesses in comparison to a direct connection to a database. 

The one exception to this is the new SSAS connector – it actually connects in real time and queries your cubes directly instead of caching the data in the cloud.

In the free version, Power BI is limited to Daily and 10K rows per hour.  This means that free is going to be limited to visualizing pre-canned excel files, already aggregated pivot tables, etc. because a 10K limit is quite small. 

For the pro version, the refresh limit is 1 hour and 1 million rows / per hour.  This also seems pretty limiting at first glance.  Assuming that there isn’t another more enterprise option, this will mean some thinking about how you design the SSAS cubes, Hive tables or other repositories so that you’re not sending over millions of rows per hour to be aggregated by your business intelligence layer.  It’s not quite clear whether a query against SalesForce for example requires a refresh or whether it’s a direct connection to the repository but assuming it’s similar to current Power Query / Power Pivot then in most cases it’s a full refresh of the data (In the current version of Power Query / Power Pivot, data refreshes are full with no support for incremental updates – if you pull in 10 million rows into your model it will refresh all 10 million every refresh).

You’ll need to do some level of aggregation before you send over the data.

It’s all About the Connectors

One of the strengths of products such as DOMO and Tableau are the large numbers of data connectors.  As organizations move to the cloud, the number of repositories of data is growing exponentially so an enterprise BI product needs to be able to connect and pull data not just from traditional data sources such as databases but also cloud services such as SalesForce, Twitter, Facebook, Google Analytics, etc.  They also need to be able to connect to NoSQL data sources such as Hadoop.

Microsoft has made some progress in expanding the number of data sources available through various updates to Power Query in the past year.  The current list includes a number of standard databases, NoSQL data sources, Azure Services and HDInsight.  Expect this to accelerate with the new Power BI service.  Currently, they have announced connectors to GitHub, Marketo, Dynamics CRM, SalesForce, SendGrid and Zendesk. 

The Weakest Link Still Seems to be the Data Management Gateway

Microsoft has released both a Data Management Gateway and a SSAS connector to enable hybrid connectivity from Power BI back to your on premise data sources.

These connectors act as a proxy between Power BI and your on premise data sources.  In the current architecture, these are applications that run in your local environment on a client computer or server.

There are some significant weaknesses to the current approach to these connectors – they are really just client applications, not robust service applications.   At least in their current incarnation, I wouldn’t consider them enterprise class for a number of reasons:

  • Power BI Analysis Services Connector can only connect to SQL Server Analysis Services Tabular Models – no support for multidimensional models yet.
  • The Power BI Analysis Services Connector has to be continually running but it runs as a desktop client.  There is no clear way to install it as a Windows Service.  There is also no clear way to run it with redundancy or to load balance multiple servers to provide high availability.
  • You cannot run the Power BI Analysis Services Connector and the Data Management Gateway on the same computer.  That means two separate servers are required at the moment to handle different proxy mechanisms.
  • The error checking, exception handling and traceability when gateway refreshes happen is not very robust and not easily monitored outside of Office 365 (e.g. by System Center or other aggregate monitoring tool)
  • Credentials are based on a single refresh account and not by the logged in user.  Refreshed data is available to all users who have access to the report.

Architecture