XML for Analysis (XMLA)

Microsoft SQL Server 2008 Analysis Services Unleashed Review Chapter 32 

XMLA (sometimes abbreviated XML/A) is a standard industry protocol for accessing multidimensional data.  It was created to be used with the stateless HTTP protocol, and therefore represents a complete transaction within one message (therefore independent of previous or subsequent communication).

To me, an important feature of XMLA is allowing for sessions, a server-side state devoted to a particular user.  Sessions require initial authorization and authentication, and have an expiration.  To analysts, having a session means having the ability to make temporary or permanent changes in Analysis Services based on some user change.  This technology makes Visual Analytics possible (where one could change a graph or other visual and change the underlying data).  The book talks about “what-if” scenarios (page 580) where data could be temporarily changed on the server.  That ability is also important since many cubes may be large, and a local cube may be prohibitive or impossible.  Increasingly, I believe that concepts like Visual Analytics are going to push OLAP technologies to be more of a two-way communication between analysts and data sources.  As I said in the last chapter review, creating a data mining structure or model changes the Analysis Services database.  Perhaps skilled analysts are the missing data source sometimes overlooked when business intelligence architects design systems.  As I have promoted, I believe data mining differs from pure machine learning since the activity requires interaction with people, and not just pure computational mathematics and performance.

The MSDN Documentation provides a topic, XML for Analysis Reference (XMLA).  The chapter has some examples of applied XMLA and therefore provides insight into how Analysis Services uses this technology.

Gorbach, I., Berger, A., & Melomed, E. (2009). Microsoft SQL Server 2008 Analysis Services Unleashed. Indianapolis, IN: Pearson Education Inc.
ISBN: 0-672-33001-6