While optimistic, Nakahara refused to be drawn into a discussion on when this could be achieved. Although Fujitsu is among the top five server makers in the world, there is currently a huge gap between the top three players and the rest of the chasing pack.
Still, early indicators point to an interesting battle with titans like HP, IBM and Sun Microsystems.
Fujitsus weapon of choice? Its Solaris-based PRIMEPOWER line of servers, which the company is positioning as the most robust, scalable and reliable in the world, Nakahara said.
Earlier this year, Fujitsu refreshed its PRIMEPOWER server line with the release of four low- and mid-range systems that boast performance improvements of as much as 40 percent over previous products.
The four models—PRIMEPOWER 250, 450, 650 and 850—scale from two to 16 processors, and come with a range of autonomic and high availability features that complement the high performance GHz class SPARC64 V processors.
According to Nakahara, servers these days no longer just provide internal business support. Instead, they must operate 24x7 at minimum cost and provide high levels of usage flexibility.
Like its flagship PRIMEPOWER 2500 (which contains up to 128 CPUs and 320 PCI connections), the newly introduced models incorporate innovations like automatic error detection and self-correction initially developed and tested for large-scale mission-critical areas of utility computing and other 24x7 environments.
These servers maximize uptime and throughput by tapping on hardware Reliability, Availability and Serviceability (RAS) features such as fully redundant and hot-swappable hardware components, as well as RAS on a chip—a capability that is more common on mainframes, according to Nakahara.
"As customer performance demands continue to increase at the entry level, Fujitsus PRIMEPOWER 250 and 450 are addressing the most urgent computing needs by giving customers mainframe-like capabilities, reliability, and availability," said Vernon Turner, group vice president of Global Enterprise Server Solutions at IDC.
To minimize unscheduled downtime, PRIMEPOWER servers are designed to automatically reboot whenever there is a fault in the CPU, memory, or PCI bus. The RAS on a chip capability helps by retrying a failed instruction at the hardware level rather than executing it 17 times before moving to the CPU level to check for a problem (as is typical in the Solaris operating system). Hence, it does not need to flush and rebuild the software buffers before re-executing the failed instruction.
Furthermore, the servers can be monitored remotely via the Web, e-mail or a private connection for very fast turnaround on preventive maintenance, component replacement and troubleshooting support.
Hoping to grab a bigger share of the server market, Fujitsu has announced a partnership with Intel to build larger systems, including Itanium-based servers that will scale up to 128 processors by 2005, according to company officials. These systems will be targeted at enterprises looking to consolidate their applications onto fewer servers.