In 1985, Microsoft and IBM announced "a long- term joint development agreement for development of operating systems and other systems software products." This announcement was the beginning of OS/2, a successor to the Microsoft MS-DOS operating system. OS/2 would be a more complete and robust operating system. It would exploit the powerful new personal computers based on the Intel 80286 processor. It would allow multitasking applications, each with its own address space and each running in the safe ring 3 of the Intel four-ring protection scheme of the 80286. Machines sold in 1986 would be vastly more powerful than the original IBM PC (an Intel 8088-based machine) of just a couple years earlier, and OS/2 would signal a new age in harnessing this power. That was the plan.
OS/2 was formally announced in April 1987, with shipment promised by the end of the year. (OS/2 version 1.0 was released to manufacturing on December 16, 1987.) But shortly after the joint declaration, IBM announced a special higher-end version of OS/2 called OS/2 Extended Edition. This more powerful version would include the base OS/2 operating system plus an SQL database called OS/2 Database Manager. OS/2 Database Manager would be useful for small applications and would be partially compatible (although not initially interoperable) with DB/2, IBM's flagship MVS mainframe database, and with the lesser-used SQL/DS, which ran on slightly smaller mainframe computers using the VM or VSE operating systems. OS/2 Database Manager would also include Systems Network Architecture (SNA) communications services, called OS/2 Communications Manager. As part of its sweeping System Application Architecture (SAA), IBM promised to make the products work well together in the future. (Database Manager later evolved into today's DB2/2.)
But if IBM could offer a more complete OS/2 solution, who would buy Microsoft OS/2? Clearly, Microsoft needed to come up with an answer to this question.
In 1986, Microsoft was a mere $197-million-per-year business, with 1153 employees. (Ten years later, Microsoft had revenues of nearly $6 billion, with almost 18,000 employees .) Microsoft's products were entirely desktop focused, and the main bread-and-butter product was MS-DOS. Client/server computing was not yet in the vernacular of Microsoft or the computer industry. Data management on PCs was in its infancy. Most people who kept data on their PCs used the wildly popular Lotus 1-2-3 spreadsheet application to keep lists (although many were discovering the limitations of doing so). Ashton-Tate's dBASE products (dBASE II and the recently released dBASE III) had also become popular. Although a few other products existed, such as MicroRim's Rbase and a relatively new product from Ansa Software called Paradox, Ashton-Tate was clearly king of the PC data products. In 1986, Microsoft had no database management products. (Beginning in 1992, Microsoft would go on to tremendous success in the desktop database market with Microsoft Access and Microsoft FoxPro.)
IBM's Database Manager wasn't in the same category as products such as dBASE, Paradox, and Rbase. Database Manager was built to be a full-fledged, blood-and-guts database (with atomic transactions and a full SQL query processor), more similar to traditional minicomputer-oriented or mainframe- oriented systems such as IBM's DB/2, or Oracle, or Informix. Microsoft needed a database management system (DBMS) product of the same caliber, and it needed one soon.
Microsoft turned to Sybase, Inc., an upstart in the DBMS market. Sybase hadn't yet shipped the first commercial version of its DataServer product (which it would do in May 1987 for Sun workstations running UNIX). Although certainly not a mainstream product, the prerelease version of DataServer had earned a good reputation for delivering innovative new capabilities, such as stored procedures and triggers, and because it had been designed for a new paradigm in computing: client/server.
As is true in all good business exchanges, the deal between the two companies was a win-win situation. Microsoft would get exclusive rights to the DataServer product for OS/2 and all other Microsoft-developed operating systems. Besides getting royalties from Microsoft, Sybase would get credibility from Microsoft's endorsement of its technology. Even more importantly, Sybase would gain a beachhead among the anticipated huge number of personal computers that would be running the new OS/2 operating system.
Because the transaction-processing throughput of these OS/2 systems wasn't expected to be high, Sybase could use the systems to seed the market for future sales on the more powerful UNIX system. Microsoft would market the product in higher volumes than Sybase could; it simply wasn't economically feasible for Sybase's direct sales force to deliver what would essentially be the first shrink-wrapped release of a full-fledged, blood-and-guts database to PC customers. Higher volumes would help Sybase win more business on its UNIX and VMS platforms. On March 27, 1987, Microsoft president Jon Shirley and Sybase cofounder and president Mark Hoffman signed the deal.
In the PC database world, Ashton-Tate's dBASE still had the reputation and the lion's share of the market, even if dBASE and Sybase DataServer offered capabilities that were extremely different. To gain acceptance, this new, higher-capability database management system from Microsoft (licensed from Sybase) would need to appeal to the large dBASE community. The most direct way to do that, of course, was to get Ashton-Tate to endorse the product so Microsoft worked out a deal with Ashton-Tate to do just that.
In 1988, a new product was announced with the somewhat clumsy name Ashton-Tate/Microsoft SQL Server . Although not appearing in the product's title, Sybase was prominent in the product's accompanying information. This new product would be a port of Sybase DataServer to OS/2, marketed by both Ashton-Tate and Microsoft. Ashton-Tate had pledged that its much anticipated dBASE IV would also be available in a Server Edition that would use the dBASE IV development tools and language as a client to develop applications (for example, order-entry forms) that would store the data in the new SQL Server product. This new client/server capability promised to give dBASE new levels of power to support more than the handful of concurrent users that could be supported by its existing file-sharing architecture.
Ashton-Tate, Microsoft, and Sybase worked together to debut SQL Server on OS/2. (This was the first use of the name SQL Server . Sybase later renamed its DataServer product for UNIX and VMS Sybase SQL Server . Today Sybase's database server is known as Sybase Adaptive Server .)
The first beta version of Ashton-Tate/Microsoft SQL Server shipped in the fall of 1988. Microsoft made this prerelease version available at nominal cost to developers who wanted to get a head start on learning, developing for, or evaluating this new product. It shipped in a bundle known as the NDK (network development kit) that included all the software components needed (provided that you were developing in C) to get a head start building networked client/server applications. It included prerelease versions of SQL Server, Microsoft LAN Manager, and OS/2 1.0.