After a late 1993 decision to become a mainstream database management system vendor, Microsoft began work on its first in-housedeveloped version of SQL Server. Prior to that point, Microsoft was a porting partner for Sybase SQL Server, adapting it to Microsoft operating systems and enhancing it with tools more suitable for the workgroup and departmental environments that used Microsoft-based servers. In 1995 Microsoft shipped SQL Server 6.0, and followed it up with SQL Server 6.5 in 1996. These two releases, developed entirely in-house, based on the Sybase architecture and code, allowed Microsoft to push beyond its original workgroup and departmental markets into some enterprise applications. But even before SQL Server 6.0 shipped, work had begun on a revolutionary new SQL Server. A new query processor, cursor engine, database API (OLE DB), and other database management system components were already under development. And initial discussions to acquire OLAP technology had already taken place.
SQL Server's transformation was planned as three releases. The first release would focus on the basic re-architecture of SQL Server and making it suitable for enterprise use. The second release would be a quick-turnaround release to address key customer concerns and plug any critical competitive holes that had emerged. And the third release would be the one where Microsoft could fully exploit the new architecture to address a broad range of new customer requirements.
With the first release, SQL Server 7.0, Microsoft successfully ripped out the internals it had inherited from Sybase and replaced them with the new technology it had developed in-house. This approach, rather than building a new database system completely from scratch, was employed to retain compatibility for the thousands of applications already running on SQL Server. Customers experienced major improvements in reliability, scalability, performance, and manageability. SQL Server also became the first database product to incorporate all the tools necessary for building data warehouses when it included both extraction, transformation, and loading (ETL) tools and an online analytical processing (OLAP) engine. With its Distributed Query feature and OLE DB architecture, it also became the first database system to make heterogeneous data access a standard integrated capability. Replication, a SQL Server strong-suit since version 6.0, underwent a major upgrade in capabilities. Perhaps most importantly, SQL Server 7.0 focused on automating or eliminating many of the tuning and management demands that had historically been required of database systems.
The second release, SQL Server 2000, was a refinement of SQL Server 7.0. XML was just emerging as an interesting topic in computing, and the SQL Server team became convinced that adding support for XML was critical. Materialized (a.k.a. indexed) views, originally envisioned as a feature for the third release, had become a hot requirement as a result of the then-popular TPC-D benchmark. And numerous modest performance, scalability, and availability improvements were incorporated to enable SQL Server to continue to meet enterprise requirements. Subsequently to the initial release of SQL Server 2000, SQL Server was enhanced with better support for mainframe-class servers such as the HP Superdome, Unisys ES/7000, and 64-bit Intel Itanium-based servers; a major security update; new XML capabilities; and new Reporting Services and Notification Services modules. The new Reporting Services feature, along with major OLAP upgrades and new data mining capabilities included in the initial SQL Server 2000 release, and various server upgrades such as materialized views, made SQL Server 2000 a leading data warehousing platform.
Work on SQL Server 2005 began late in 2000, with a goal of delivering on most of the remaining capabilities envisioned in the original three-release plan. Leading the priorities, the original vision of "Visual Basic stored procedures" had evolved to be full support for the Microsoft .NET CLR (and thus all CLR languages). As anticipated with SQL Server 2000, XML had indeed become a major force in the application development world, and it would become core to nearly all new work in SQL Server 2005. Along with these technology shifts, new features such as direct HTTP access and subsystems such as Service Broker would enable a greatly enhanced SQL Server application model. On the data warehousing front, Analysis Services would receive a major overhaul, and SQL Server's Data Transformation Services would be replaced by a new enterprise-class ETL tool. At the same time, nearly every other area of SQL Server would receive major upgrades.