Will Embedded BI Put an End to Data Warehousing?

I say a big "NO". Aside from the fact that many organizations are already tied to their data warehousing infrastructure, companies seem to be leaning towards implementing data warehouses with the goal of creating a more holistic view of the organization, while at the same time maintaining historical data for trends analysis. It seems as if there will continue to be room for both embedded BI and data warehousing within organizations having to choose one over the other.

Data Warehousing, as we know it (defined as the data repository from which reporting or analytics is done), being the sole enabler of BI is already passe. It is clear that Data Management in its broadest sense is fundamental to BI. Given the growth of data in enterprises, 'Divide & Conquer' is a good data management strategy, where data assets in an organization are managed based on how information is to be extracted & used. To me, embedded BI is one way of managing data and associated information, where BI blends seamlessly with the application itself. I foresee the usage of embedded BI primarily for operational analytics while data warehousing and other forms of data management will continue to play its role for strategic BI.

Oh! By the way, the answer to the forum question is NO!

In the context of this question, the recent announcement by SAP of its Business Analytic Engine (BAE), makes interesting reading! Is there a connect? Does anyone else see it or is it just that I am imagining things here?

The answer is "NO". Embedded BI can get rid of localized reporting(cannot call them datamarts) or at least bring some sanity to that. DW will evolve with time, but the starting point of DW is enterprise view and embedded BI cannot provide that. Embedded BI can help improve ease of implementation of BAM. Embedded BI and DW can serve two different purposes and will evolve differently.

Clearly this is a loaded question as many information management professionals will almost instinctively answer this question as a NO. Let’s clarify here, embedded Business Intelligence is really about ISV’s OEM’ng business intelligence products such as Jaspersoft into their product. The only thing it has to do with data warehousing is if the ISV requires a database in the context of embedding a business intelligence solution.
Lets step back for a moment. Many large software companies including SAP, who have acquired Business Objects and soon Sybase or IBM who has acquired Cognos try to sell their respective database products alongside their embedded business intelligence solution. ISV’s frequently prefer to separate the database decision from the business intelligence tool decision. Therefore, this really has nothing to do with putting an end to data warehousing. Consolidation in the business intelligence space will continue and data warehouses will live on.

Everyone agrees the answer is "no," but what I'd add to the thread, and perhaps is the seed of today's question, is that today's BI technologies do permit companies to skip the datawarehouse step in some cases. Specifically, data mashup technologies allow the BI application to access multiple data sources directly, and those data sources could be non-relational ones like Web services or spreadsheets. So whereas in the past, the data warehouse was a necessary step to combine all those data types and make them readable by a reporting and charting application, now a BI application can perform that role of aggregation and re-formatting.

The answer is "no" since by definition embedded BI can only utilize data that is held within the application itself. Unless the company has a homogenous application set from a single vendor (unlikely) it would be very difficult to get obtain an holistic perspective across a complete business process.

I agree with Mark that the traditional warehouse approach is not however a requirement. I think it is a little more complex than simply mashing data together wherever it lies since detailed analysis usually requires a multi-dimensional datastore. However this datastore should operate in real-time without requiring periodic cube crunches and should be accessable by intuitive, interactive visualization and analytics suitable for use by everyone making daily decisions and not just by business analysts