Microsoft offers several Data Warehouse solutions.   There is SQL 2008 R2 based solution that is the traditional one where the customer designs and deploys the DW solution using hardware of their choosing.  Another option is the Fast Track Architecture which is a predefined and tested solution.  The third option is the Data Warehouse Appliance that will release later this year called the Parallel Data Warehouse (PDW).   This last option is referred to as an appliance which means that the hardware, software and service components are defined and built. 

Having choices is always good, but that can lead to questions as to which option is best suited for a given environment.   The Fast Track Architecture and Parallel Data Warehouse are geared for the relational side of the DW.  They focus on the fact and dimension tables.  They can serve as the source for the Analysis Services where aggregated data based on the fact and dimensional tables are stored.  The first option can be used for either relational and analysis datasets.

To further differentiate between the latter two options, Fast Track Architecture is essentially a set of  guidelines for designing a balanced DW system.  The Fast Track Architecture guidance is built on best practices for data layout and loading and reference hardware configurations.  Parallel Data Warehouse is defined appliance and not just a reference set of configuration and hardware guidelines.   When selecting a Fast Track Architecture, the designer needs to know the size and expected growth of the DW to ensure the proper selection.  One cannot simply add more storage and accommodate for the growth.  The Fast Track Architecture is predicated on being a balanced system which means IO is balanced across all components.  Changing the storage would alter this balance, and the system would no longer be  a Fast Track Architecture system.    For now, the Fast Track Architecture configuration can support up to  48TB. Parallel Data Warehouse is designed for systems in the 100TBs.  PDW is extensive in that you are not set to a prescribed size,  you can also extend it by adding additional 100 TBs data racks.  The incremental size is 100 TBs. 

More info on this can be found at: