INDUSTRY

Liquid Computing Corp., developer of a new class of scalable computing systems for high performance computing, announced today the availability of the Alpha version of its LiquidIQ server, the industry's first Interconnect Driven Server. Today's computing systems are slowed down by system bottlenecks caused by disaggregated communications equipment and system control software. Industry users and government researchers have an increased demand for more processing resources, higher bandwidth, lower latency, more memory and I/O channels, but all at an optimized price and performance level. Liquid Computing has recognized that organizations are faced with cluster configurations that cannot cost effectively scale due to inefficiencies within the underlying communications networks. Liquid Computing has designed a server that far exceeds the performance and scalability levels of legacy vector machines and today's commodity clusters but at a dramatically lower cost. LiquidIQ delivers a set of managed computing and communications resources that can be configured with software commands into one or several cluster configurations, shared memory or cache coherent server regions. Industry analysts have observed a dramatic increase in smaller cluster configurations in the high performance computing (HPC) market. This deployment model has proven to be ineffective as it raises a set of management challenges, duplicates tasks and lowers the overall productivity of deployed processors inside an organization. Liquid Computing is enabling users to experience the benefits of high performance computing without compromise. "The move to a consolidated server module is inevitable as the demand for servers that enable cost-effective, faster, more responsive and customized computing continues to increase. Other markets such as storage and voice have already witnessed the benefits of consolidating redundant technologies," said Brian Hurley, CEO of Liquid Computing. "Liquid Computing is offering companies the opportunity to achieve higher scalability, performance and productivity levels that were previously not affordable to them. Our servers will change the way computing systems are designed." This underlying convergence of computing and communications resources requires several requisite technologies and services that form the basis of the Interconnect Driven Server (IDS) architecture. These include: - A fault tolerant interconnect - Virtualization & quality of service control - Autonomic control services - Policy driven resource management - Extensibility framework The sum of these requisite elements allows users to instantly marshal computing, memory, IPC and I/O resources into highly available and scalable servers that dynamically change to deliver any computing or communications task at best economics. The Alpha version of LiquidIQ is currently going through a set of tests with several industry performance benchmarks and hands on field-testing to be announced shortly. "We anticipate Liquid Computing could provide us with a single system platform to address our HPC and enterprise computing challenges," said Mark Hargrove, senior vice president of operations and CIO of DigitalGlobe. "We are pleased with the progress we have already seen and look forward to witnessing performance results under load early next year." High performance computing (HPC) systems are positioned for long-term growth and the market is expected to grow 6.5% annually over the next five years (IDC). The enterprise market is also expected to make a shift towards a consolidated model. Liquid Computing is converging computing and communications networks to introduce a disruptive change in the HPC and enterprise markets. The company is addressing these markets' key communications and control challenges with a scalable computing solution. Supercomputing 2005 Liquid Computing will be featuring the alpha version of LiquidIQ at Supercomputing 2005 this November in Seattle http://sc05.supercomputing.org/ . The demonstration unit will also feature third party simulation software. Liquid Computing's CTO Mike Kemp will be giving a presentation entitled "HPC Developers Beware: Know Thy Networks" on Tuesday November 15 at 10:30 a.m. in room 617.