Redmond Diary

By Andrew J. Brust

Blog archive

BI (Up) in the Cloud

In the new film “Up in the Air,” George Clooney's character, Ryan Bingham, poses a question relevant to the premise of BI as a cloud computing offering. Bingham is a career transition specialist -- i.e. someone who fires people as an outsourced service – and he insists that his services must be delivered on-premise (if you will), despite his firm's new initiative to start doing so via Web conference technology. That initiative is being pushed by his colleague, Natalie Keener (played by Anna Kendrick), a new hire and recent graduate. The film explores this inter-generational debate and competition between the effectiveness of personal presence and the efficiency of the Internet.

A similar debate is playing out between BI startup companies, many of which are cloud pure-plays, and the traditional BI heavyweights. The startups, with names like Oco, ParAccel, and Birst, insist that the efficiency, ease of provisioning and on-demand scale of the cloud is exactly what's needed to move BI more into the mainstream. Meanwhile, the traditional BI players like IBM/Cognos, Oracle/Hyperion, SAP/BusinessObjects and even Microsoft don't necessarily agree, and don't see sufficient customer demand to warrant investment in the new approach.

If Ryan Bingham were a BI specialist, he would argue (and rightly so), that enterprise BI implementations vary greatly, require significant business analysis and sometimes even business process reengineering, both which are very interactive, high-touch services best delivered in person. Even after implementation, the volumes of data, the processor-intensive ETL (extract, transform and load) jobs, and the data- and query-specificity of cache, index and cube optimizations don't lend themselves well to a multi-tenant, hosted environment.

Natalie Keener would likely counter (and not without merit) that in-memory database technology can render many of these post-implementation concerns moot. And I bet she'd also argue, even though it would incense Ryan, that with enough customer volume, commonalities in the analysis and business process reengineering delivery would start to emerge, making the services more mappable, repeatable and deliverable by new trainees.

Who's right? For large enterprises, I'd say Bingham's right, today. For smaller organizations looking for analytics on their Web logs, CRM/ERP transactions and other common data repositories, Natalie's vision may be feasible quite soon. Even larger organizations with more customized needs might fit her paradigm in the not-so-distant future. Only issue is that the technology's not ready for them yet, and even if it were, the organizations would not be ready for the new technology.

As a case in point, take a look at Microsoft's BI stack, cloud stack and SaaS stack. Microsoft has a lot of cloud BI pieces already in place. It also has some more work to do. SharePoint is key: its ties to Excel Services and the Excel Web App make it the sensible end-user BI point of delivery. Then take a look at PowerPivot, which uses the high-compression, in-memory technology Natalie would love. Its server engine integrates with SharePoint, which can initiate and load-balance instances of the server, according to demand, which is very cloud-like indeed. But will SharePoint Online support all of these products and features? It's hard to tell. And what about SQL Azure? Seems like a great on-demand data warehousing platform. Until you start to contemplate its 10GB limit in database size!

So Microsoft isn't there yet; Bingham would be vindicated. But imagine If the Excel Web App and Excel Services, which are already kissing cousins, merged. And imagine if SQL Azure databases could be larger, and a scale-out technology, like that in the forthcoming SQL Server “Madison” product, were used to aggregate a collection of such databases into a single data warehouse. Now, finally, envision PowerPivot running as part of SharePoint Online and in the core Analysis Services product, which would become a feature within SQL Azure. With all of this, Microsoft would really have something for cloud BI. And, by then, their customers might just be ready.

On that day, Natalie might feel smug, like she were right all along. But could she last that long? One cloud BI startup, LucidEra, has already folded. And by the time such product and customer maturity were reached, Ryan may be very comfortable with the change, gracefully delivering his old services within it. Granted, he, like Microsoft, would be older. But if the offering's a good one, then I bet it wins. George Clooney, and his admirers, would corroborate that, I imagine.

Posted by Andrew J. Brust on 12/31/2009 at 1:15 PM


comments powered by Disqus

Featured

  • Color Shards

    Sharing Data and Splitting Components in Blazor

    ASP.NET Core Version 3.1 has at least two major changes that you'll want to take advantage of. Well, Peter thinks you will. Depending on your background, your response to one of them may be a resounding “meh.”

  • How to Create a Machine Learning Decision Tree Classifier Using C#

    After earlier explaining how to compute disorder and split data in his exploration of machine learning decision tree classifiers, resident data scientist Dr. James McCaffrey of Microsoft Research now shows how to use the splitting and disorder code to create a working decision tree classifier.

  • Microsoft: Move from Traditional ASP.NET to 'Core' Requires 'Heavy Lifting'

    There are plenty of reasons to move traditional ASP.NET web apps -- part of the old .NET Framework -- to the new cross-platform direction, ASP.NET Core, but beware it will require some "heavy lifting," Microsoft says.

  • Purple Blue Nebula Graphic

    How to Compute Disorder for Machine Learning Decision Trees Using C#

    Using a decision tree classifier from a machine learning library is often awkward because it usually must be customized and library decision trees have many complex supporting functions, says resident data scientist Dr. James McCaffrey, so when he needs a decision tree classifier, he always creates one from scratch. Here's how.

  • Blazor's Future: gRPC Is Key

    Blazor guru Steve Sanderson detailed what Microsoft is thinking about the future of the revolutionary project that enables .NET-based web development using C# instead of JavaScript, explaining how gRPC is key, along with a new way of testing and a scheme for installable desktop apps.

.NET Insight

Sign up for our newsletter.

Terms and Privacy Policy consent

I agree to this site's Privacy Policy.

Upcoming Events