Posted by jj as Database
Unisys and Microsoft collaborated on a proof of concept to demonstrate that the Microsoft SQL Server 2005 product suite will support even the most demanding enterprise business intelligence initiatives. The proof of concept shows that SQL Server 2005 can deliver the same or better results for a high-performance enterprise-class data warehouse than the more-costly alternatives.
We delivered an optimized data warehouse solution using SQL Server 2005, SQL Server 2005 Integration Services, and SQL Server 2005 Analysis Services on a Unisys ES7000/one Enterprise Server with x64 processors. We conducted a series of tests for extraction, transformation, and load (ETL) operations and online analytical processing (OLAP) cube building. These tests measured throughput, scalability, and performance as the amount of data and execution parallelism were increased. SQL Server 2005 demonstrated the ability to use all available CPUs and memory for both the bulk data load and cube processing testing. In this proof of concept we outperformed the direct competition (Oracle-based solutions) by at least 45 percent. The test results documented in this paper prove that the SQL Server 2005 suite running on the Unisys ES7000 server can take advantage of parallelism and multiple processors to deliver extremely high rates of ETL and OLAP cube processing throughputs. Our team of 12 specialists from Unisys, Microsoft (U.S. and Germany), and Platon implemented this fully-functional data warehouse in less than three weeks.
Download pdf Data Warehouse Performance: Unisys and Microsoft Achieve Record-setting Benchmark