Brief History

Client-server systems began to emerge in the United States in the early 1980s as computing transitioned from large mainframes to distributed processing using multiple workstations or personal computers. Corporations quickly adopted client-server systems, which became the backbones of their office automation and communication infrastructure.

Ken Thompson and Dennis Ritchie working at Bell Labs developed UNIX in 1969 as an operating system that ran on minicomputers. UNIX soon became widely used on minicomputers. Workstations running UNIX operating systems had been around since the early 1980s, but Japanese corporations only started developing and marketing UNIX workstations in the second half of the 1980s. At the beginning of the 1990s, workstation market growth tailed off due to the increasing power and performance of personal computers, and by the mid-1990s the workstation market was in decline. Initially, only some high-end workstation models were utilized as servers, but over time this became the prime application of UNIX workstations.

Since servers were mainly targeted for business applications that involved connections with multiple clients, they not only had to deliver processing power; they also had to provide stability, fault-tolerance, and scalability on par with mainframes. Servers were constructed to support upgrades to redundant or even quadded redundant architectures and used multiple hard disks in RAID arrays. Fault-tolerant systems were also arranged with non-stop operating systems for applications requiring continuous operation. Japanese manufacturers initially developed UNIX servers using their own technology, but later they developed servers with licensed technology or sold OEM products. Some of the more common server processors included MIPS Technologies' line of RISC processors (R3000, R4400, and R10000), Sun Microsystems' SPARC series (SPARC, SuperSPARC, hyperSPARC, and UltraSPARC (a 64-bit processor)), and Hewlett-Packard’s PA-RISC processors (7000 series and 8000 series (64-bit processors)).

The introduction of UNIX System V in 1992, which supported multiple processors, sparked the development of multiprocessor architectures to boost performance. Both the eight-processor NEC UP4800/770 in 1995 and the 16-processor Hitachi 9000V/VT800 in 1997 set new industry speed records for servers. Fujitsu announced the GP7000F 2000 Model in 1999, which featured a symmetrical multiprocessor (SMP) architecture consisting of 64 SPARC 64-GP processors, followed by the PRIMEPOWER series in 2000 as the company's new global UNIX server brand. The PRIMEPOWER 2000 had a SMP architecture that supported up to 128 processors. Domestic manufacturers started out with their own proprietary versions of UNIX but all eventually switched to UNIX versions developed by U.S. vendors. At the end of the 1990s, the most commonly used UNIX operating systems in Japan were Sun Microsystems' Solaris, Hewlett-Packard's HP-UX, and IBM's AIX.

Server manufacturers utilized redundant system architectures and independent service processors to improve the reliability of their servers. For example, Hitachi's fault-tolerant FT6100 server in 1991 featured processor boards with triply redundant functional modules, and its 3500/730FT and 750FT models, which came out in January 1995, featured a quad processor redundancy (QPR) architecture, which had two redundant-processor systems (for a total of four processors). Toshiba's 1998 UX2000i server included hot-swappable hard disks in a RAID array as well as a system service processor that was independent of the unit's CPUs, which enabled remote server operational management and servicing via a network.