Understanding the Fast Track Data Warehouse Reference Architectures

Microsoft recently released a series of resources about its SQL Server Fast Track Data Warehouse that are excellent for anyone looking to design and implement the relational database portion of a SQL Server data warehouse.  The Fast Track reference architectures provide tested, preconfigured architectures and architectural guidance for a business intelligence (BI) solution's database components and hardware systems, taking the guesswork out of hardware and architectural decisions for the database layer of SQL Server BI solutions. The Fast Track reference architectures aren’t tailored around any particular vertical market or business application; instead, the Microsoft Fast Track home page says features of the Fast Track reference architecture aim to
• Accelerate your data warehouse roadmap with pre-tested hardware configurations
• Virtually eliminate hardware testing and reduce tuning with better Data Warehouse performance out-of-the-box
• Scale from 4 up to 32 terabytes using compression capabilities in SQL Server 2008 Enterprise
• Enjoy a lower cost of ownership through better price performance, rapid deployment and industry-standard hardware
• Choose the right performance, scalability and pricing to suit your business needs

What distinguishes a Fast Track Data Warehouse from a typical OLTP database is that all of its components—from CPU to disk—are balanced and tuned for OLAP, with a primary emphasis on eliminating potential performance bottlenecks. The architectures are designed for sequential, rather than random, I/O and are designed to be able to provide sustainable speeds up to 200 MB/s per CPU core. Microsoft claims that this level of performance offers similar performance at up to 1/5 the price of solutions from other vendors. Without a large amount of real-world data warehousing expertise it’s easy for customers to spend large amounts of money on a server environment but end up with a system that isn’t well tuned for the sequential I/O nature of data warehousing. Fast Track relies heavily on the use of preselected hardware components to ensure that artificial system bottlenecks aren’t introduced into the data warehouse environment. The approach is designed to be scalable within Madison, Microsoft’s scale out data warehouse technology based on its acquisition of DATAllegro technology. 

Dig your teeth into understanding this new reference architecture by taking advantage of the Fast Track home page. You’ll find detailed white papers that provide background on the approach as well as detailed prescriptive guidance within the Implementation guide. It’s worthwhile reading for anyone involved in the ownership, management, planning, or design of a BI solution. In addition, the SQL Server Fast Track Data Warehouse webinar will help you tie all of the pieces together.

Fast Track doesn’t solve all of your data warehouse problems or magically build your data warehouse, but it will provide very valuable information to ensure that your final solution will scale in the short term and long term.