Parallel Data Warehouse Appliances Ship

SQL Server 2008 R2 Parallel Data Warehouse (PDW) is an appliance platform for data warehouses (special-purpose databases that store large volumes of data for reporting, data mining, and other functions to aid decision making in organizations). Based on SQL Server 2008, the platform is intended for data warehouses that grow into the hundreds of terabytes, spread over large numbers of low-cost servers, to support business intelligence (BI) needs. PDW released to manufacturing in Nov. 2010, and systems will be available from OEM partners in Dec. 2010 and throughout 2011.

Cost-Efficient Scaling

PDW was designed to provide a low-cost, easy-to-deploy appliance by building it as a preconfigured system built with commodity components. The PDW platform (formerly code-named Project Madison) is actually based on SQL Server 2008 (not 2008 R2) as well as technology acquired with Datallegro in July 2008.

Rather than a single system with a large number of processors (“scale-up”), PDW delivers “scale-out” architecture by using multiple lower-cost servers (nodes) installed into two to four data center racks, depending on the size of the system ordered. For example, HP PDW systems provide servers with two physical processors (CPUs), with six, four, or two logical processors each, with their own low-cost local storage. PDW systems can be ordered with a minimal number of nodes, and they can easily be expanded later to support up to 40 nodes and 1 petabyte of data without requiring a massive upgrade effort.