At SC2000, Sun has demonstrated the products, services and technologies which comprise a virtual supercomputing farm, thus allowing customers to harness unused resources, such as CPUs, memory and storage across the network to handle the most data-intensive applications. "It is clear that the future of supercomputing is network-centric, and lies in both the shared power of smaller, parallel Web-serving systems and large, co-operating clusters of Shared Memory Processing (SMP) servers", stated Steve Campbell, Director of Marketing, Enterprise Systems.
"Rather than focusing solely on traditional islands of computing power in a single location, Sun is applying the principles of Net Effect in order to bring disparate commercial and scientific communities the hardware they need to share data and resources across the globe", as Mr. Campbell added. In the last year, Sun has established itself as the second-leading HPC provider, growing its market share by 20 times since the company entered the supercomputing market in 1996.
In the "World's Top 500 Supercomputers" list that was issued on November 3rd, Sun captured the second-highest number of systems with 92 entries, representing more than eighteen percent of the total list. Recent announcements illustrate Sun's continued commitment to the HPC market, and the company's expanded efforts to tackle new generations of commercial applications such as decision support, data mining and bioinformatics.
Last September, the company introduced the Sun Discovery Bioinformatics Programme, that enables biotechnology leaders to develop highly synergistic and interoperable solutions required to power the discovery pipeline. In addition to aligning themselves with Sun technologies, programme members will also receive technical support in optimising and benchmarking, as well as developing and deploying solutions throughout networked computing environments. The company equally presented the Sun Informatics Advisory Council, a consultative group whose commercial and academic members will assist in mapping out Sun's future hardware, software and service offerings for the life science industries.
Sun and the College of William and Mary announced on November 1st the installation of one of the largest academic computing clusters worldwide. The "SciClone" HPC cluster is actually a grid of four sub-clusters that consists of 160 processors and offers a theoretical peak performance of 115 gigaflops, which is the equivalent of 115 billion mathematical operations per second. The grid-like environment will allow faculty and students to research the issues facing users of larger distributed systems utilising the Internet.
The Sun booth at SC2000 illustrated these achievements and more with a variety of demos and presentations, including the High Performance Storage System (HPSS) 4.2. Now available for the Solaris operating environment, IBM's HPSS 4.2 is open system software designed to manage petabytes of data and to move large data objects between supercomputers, workstation clusters, and storage libraries at speeds several times faster than today's software systems.
HPSS can manage parallel data transfers from multiple network-connected disk arrays at rates greater than 1 Gbyte per second, making it possible to access high definition digitised video in real time. This demo featured HPSS improvements for HPSS 4.2, such as the Sun port and the efficiency of remote movers. Visitors were be able to request data from local and remote sources, with data rates commensurate with the bandwidth of interfaces to the local and remote movers.
Image-guided surgery visitors were able to view first-hand how Sun's high-performance workstations and Sun Enterprise servers can help surgeons remove previously inoperable brain tumours. This is already being done at Brigham and Women's Hospital. The Surgical Planning Lab, which is part of the Department of Radiology, develops software tools to bring the power of imaging and graphics to novel areas of medicine. SPLViz is an application using Java 3D technology which displays anatomical models and cat- or magnetic resolution imaging-scans of patients to assist physicians during diagnosis and surgery.
In addition, Sun demonstrated the Grid Engine and the concept of Technical Compute Farm (TCF). With Sun's acquisition of Gridware Inc. in July 2000, and its industry-proven distributed resource management (DRM) solutions, Codine and the Global Resource Director (GRD), Sun is now offering its own suite of DRM software tools. Grid Engine workload management software was showcased at the Technical Compute Farm demo. TCF acts as a single entity on the network, which integrates Sun Enterprise servers, storage and networking to solve today's compute-intensive problems.