During TechEd 2011, Microsoft announced an important update to its BI roadmap. The reason why this is important is related to the previous announcement, which I discussed in November 2010 by including links to several sources and comments. At PASS Summit 2010 the announcement regarding the next version of SQL Server, codenamed “Denali”, caused a lot of discussions because it seemed that there would have been a “competitive” version of Analysis Services that would have been required existing projects made in UDM to be not upgradable to the “new model”. And someone wondered if the lifecycle of UDM (Unified Dimensional Model, which is run by the MOLAP engine) started a descending curve.
In these months, I’ve been involved in many discussions with the development team with other partners, influencers and MVPs. What was a perception at that time was founded on some concrete facts: there was no plan about merging these technologies or providing any type of migration paths. Many worries about a possible end of UDM was not based on reality (the number of SSAS adopters is still growing every day) but the perception of a lack of commitment by Microsoft on UDM was the real issue. Many people asked for a change in this direction, looking for a converging path between these two technologies.
Based on my experience, I know that there are BI problems that need a multidimensional approach, but there are other problems that better feet with a more “traditional” relational approach. In the middle, a very large number of problems can be solved by both approaches at the same cost and with similar performances. These two worlds were separate: each one had its own tools, languages, data models and so on. And they would have been separated also in the future, based on the understanding everyone had with the previous roadmap.
Now, Microsoft is raising the bet on its Business Intelligence platform. These two worlds will grow together and we’ll see a convergence towards a componentized approach where every technology for every layer of the solution will be able to interact with the other components. In other words, the same client tool will be able to query any model, regarding of the technology used to implement it (either relational or multidimensional). From the user perspective, this means that Excel will be able to query any relational or multidimensional model, and the same will be true for the new shining Crescent tool! From the architectural perspective, any existing investment on any technology will be preserved. For the large number of scenarios where either models can be used, you don’t have to choose between a mature technology (UDM) that is not supported by modern tools (Crescent) and a new technology (Vertipaq) that misses some advanced feature (tool dimensions?) your user might expect in the solution because it was present in other solution he used in the past.
Thus, we’ll have a single unified model, called BI Semantic Model (BISM). BISM will have two technologies to model data: the Multidimensional one (formerly known as UDM) and the Tabular one (that will be an enhancement of the current PowerPivot data model experience). You will have two languages, MDX and DAX, to define business logic and to query the model. And you will have two engines, MOLAP and Vertipaq (with their corresponding counterpart for accessing data in passthrough mode, ROLAP and Direct Query). The great and wonderful big news is that in BISM it will be possible to combine these components as you need: DAX query over a multidimensional model, MDX Script over a Tabular, and any client able to access to any model.
This is a great and ambitious vision. It might seem complex. But there is no silver bullet in this world, when you have to implement a real solution, you have to face different requirements and up to day nobody found a single technology that can be used for any needs. Implementing this vision in its completeness will require several years and its first release in Denali will not offer the full flexibility that has been promised. But the new commitment stated by Microsoft about this vision is what many partners were looking for. Their investments, and the investments made by their customer, will be preserved with this roadmap and any data model created in the past (and also today) will benefit in the future by the appearance of new tools and new products. Without requiring a complete rewrite of the solution and a complete retraining of their resources. In a competitive world and in a complex economy phase, this is definitely the most important message of this announcement.
As usual, we’ll have to wait for the concrete bits of Denali and of the following releases to see if facts confirm the words. But I have many signs confirming that also the facts are really going in that direction. I will write more technical posts about the implication of this announcement in the future, especially after I’ll be able to talk publicly about technical details that are now under NDA. In the meantime, you can also read this very interesting comment written by Chris Webb. And, of course, you can comment my post with sharing your perception – I’m interested in getting your feedback!