Freigeben über


Soooooo...you now have two approaches to hosting BI in the cloud (Power BI vs. Azure VM Role for BI Deployment)

With the announcement of Power BI at the WPC last week, Microsoft now has *two* ways it can host your BI application in a public-cloud setting. 

  • Power BI delivers self-service BI modeling, transformation, reporting, and geospatial via Office 365.
  • Of course, you can run both the traditional (SQL/SSAS/SSRS/SSIS) and self-service (Power Pivot/Power View) by provisioning capacity through the Azure VM role. 

We're still very early in the game, of course.  Azure IaaS is still pretty new, and Power BI has a few months and some rigorous testing before it's released.  So my guidance here is far from complete.

However, it does make sense to think of the scenarios where one approach may be better than another.  At present, I find it may be a stretch to see many "better-together" stories (i.e. where BI is provisioned through both Azure VMs and Office 365) - but I'm sure an exception or two will pop up.

Here's my first stab at rationalizing the choice - I'm sure the thinking will need to evolve over the coming months.

 

Consideration Power BI Azure VMs
Connectivity to on-premise data Delivered via a net-new component called the Data Management Gateway; list of supported data sources may be limited upon initial release

Supported by Azure Virtual Networks; no specific list of data sources permitted/restricted; does require on-premise firewall configuration

Support for mobility Native apps for BI (essentially delivering Power View on mobile devices); anticipated to include Windows devices, with iOS to follow shortly thereafter

 No built-in "app" at present; can always provide browser-based access to SharePoint BI capabilities

Size of PowerPivot model 250MB (projected)

 2GB (as dictated by SharePoint)

"Conventional" BI Support

No Integration Services, Analysis Services (OLAP or BISM), Master Data Services, Data Quality Services, Reporting Services (aside from Power View),

PerformancePoint Services, Visio Services

 All SQL Services (SSIS, SSMDS, SSDQS, SSRS, SSAS) can run in an Azure VM, as can PerformancePoint and Visio Services

Natural Language Query

 Available at GA

 Not immediately available

Provisioning

 Very quick setup; little involved with creating a Power BI site in Office 365

 Pre-built Azure VMs for BI workload not currently available.  Similar architecture/configuration considerations vs. private cloud deployment

Licensing

 New Power BI SKU (requiring E3 and Office Pro Plus coverage for audience) - no explicit license for SharePoint servers or SQL Server capacity

 Requires Office Pro Plus, SQL Server (Enterprise), SharePoint (servers and eCAL) - no Power BI SKU considerations.