Improved Features for Large Databases

A new mechanism for partitioning databases in SQL Server 2005 improves the performance and manageability of large databases, such as data warehouses. In addition, SQL Server 2005 enhances replication, which administrators commonly use to scale out large, heavily used databases. These enhancements, along with previously described features aimed at improving SQL Server availability, could help Microsoft wrest market share from IBM and Oracle in the enterprise database market.

Partitioning Improves Performance and Scalability

One key scalability improvement in SQL Server 2005 is better support for database partitioning, a method for scaling databases by splitting them into partitions that reside on different storage locations on the same server (local partitioning) or on separate servers (distributed partitioning).

Partitioning improves the performance and manageability of large databases in several ways. First, partitioned databases improve query performance by spreading work across multiple disk systems and taking advantage of parallel processing. For example, a common query in data warehouses and other databases used for analysis involves data aggregation (summing regional sales results for a year, for instance). In a properly designed partitioned database, much of the workload of aggregating data can be done in parallel-for instance, multiple processors, each working on a separate partition, can simultaneously execute aggregations, which are subsequently combined into a single result.