SharePoint 2013: Supporting analytics on the back end

Analytic decision support that can be delivered rapidly and flexibly is an invaluable asset.


Microsoft’s promise of out-of-the-box business intelligence functionality was long since fulfilled in SQL Server 2005, but access to that functionality has largely been confined to people who were SQL Server-proficient. Only now, withSharePoint 2013, is that functionality available on the business side, free of deep dependency on IT.Increasingly, analytics is an ad hoc tool of decision-makers and stakeholders. Strategic advantage is increasingly a thing of the past, and rapid response is the greatest strength of the enterprise. Analytic decision support that can be delivered rapidly and flexibly is an invaluable asset, and with the new SharePoint release, Microsoft has made its best offering yet, with a toolset that requires little or no IT intervention to be effective in non-technical hands.

Big data on little desktops

The catch-22 of analytics is that big data is the best fuel for powering it – more data means better results – but working with hundreds of millions of rows of information doesn’t happen neatly on the working manager’s desk. The resulting trade-off, historically, has been the ability to do limited ad hoc analytics locally with modest batches of data (via Excel), versus static analytics from big data using custom apps or third-party packages. There’s been almost no middle ground.

But the growing need in the marketplace has been big data processing power for ad hoc analytic solutions. The entire point of business analytics is to clear noise from signal in leveraging data for decision support; what constitutes noise and signal differs in big data, problem-to-problem. Excel is an ideal tool for ad hoc analysis, ubiquitous in the marketplace, and not difficult to learn; but getting the right data into it has been problematic, even via Excel Services (part of SharePoint since 2007), and getting enough data into it has been flatly impossible.

Pitching big data from the back end

The problem is two-fold on the data side: bringing data together from different sources (particularly when some of those sources are not on the Microsoft platform), and coping with the processing overhead that results from continuous massive reads. SharePoint addresses both of these problems and the 2013 release in particular opens new doors for big data in faraway places.

OData (Open Data Protocol) is now the primary data transport facility for accessing data sources both foreign and domestic in a SharePoint context. Connectivity to business data has always been available in SharePoint, but it’s been klunky and difficult to use, and completely opaque at the business user level. The introduction of OData greatly enhances ease-of-use, and allows the business user to pull from relevant sources with minimal dependence on IT.

There’s more to the back-end picture, however. A common problem with ad hoc analytics is that the primary data source is usually a sprawling data warehouse that is neither optimized for analytics nor organized in such a way that typical business users can dig out what they need without substantial training.


PerformancePoint implementation

SharePoint 2013’s leg up on this point is its PerformancePoint implementation, which is deeper and more convenient (Kerberos delegation, for instance, is no longer required). If the enterprise is best served by a primary body of data, a separate repository for that data is in order, so that heavy hits on the source will not impact non-analytical querying.

Finally, there’s the problem of getting the right data to the right user for the right purpose. The implementation of analytics-dedicated data sources creates an opportunity for the deployment of data models that are specifically optimized for the analytics users in the enterprise, data models that offer both efficiency and flexibility in ad hoc selection of information for problem-solving. It’s true that the creation of such data models is generally beyond the typical business user; but that’s all the more reason to get all the consumers of analytics together to work out the most useful design (which IT can readily implement), to ensure maximum utility for the greatest number of users.