Electrical computers and digital data processing systems: input/ – Intrasystem connection – Bus interface architecture
Reexamination Certificate
2001-02-05
2004-05-04
Ray, Gopal C. (Department: 2181)
Electrical computers and digital data processing systems: input/
Intrasystem connection
Bus interface architecture
C710S100000, C709S208000, C370S901000
Reexamination Certificate
active
06732215
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to parallel data processing systems, and more particularly to, a unique expandable multiprocessor computer system incorporating a master computer, control computers and servant computers housed in modular units.
The advent and progression of the information age has caused a proliferation of data. Currently, many medium to large companies require document annotation, markup, workflow projections, configuration management, text retrieval, optical character retrieval, form handling, and other procedures. The hugh amount of data compiled and the management of it has resulted in two significant problems in this sector. Namely, insufficient processing capabilities and, as evident in the communications sector, a limitation on the amount of bandwidth available to move the data swiftly.
Space is a premium in corporate and ISP data centers. Compute density of board level products is therefore a key consideration when small-footprint, standard-form-factor systems are required to maximize available cabinet space. Accordingly, there has been an increasing demand for data processing systems incorporating “scalable architecture”. As this term implies, the system can grow as the need for more data processing grows. Corporations and ISP's can dynamically link up additional TI Thin Servers in a single cluster to handle enormous volumes of network transactions. Automated, pre-installed load balancing and management applications enable clusters to perform efficiently with minimum down time.
High performance computing (HPC) is a technical term describing the method in which significant computational power accesses or processes very large amounts of data very quickly, or a computing characteristic where there is a need to operate interactively across a geographically distributed network. HPC is manifested in such task s as optimization of industrial processes, computational modeling, on-line transaction processing, data mining and decision support, and complex visualization and virtual reality.
Until recently, HPC was dominated by large, specialized supercomputers.
Traditional markets for HPC systems are aerospace, automotive, chemical, pharmaceutical, environmental, government, petroleum, and a wide range of others involving electromagnetics, chemistry, fluid dynamics, weather prediction and
3
D seismic processing. Easily recognizable users within these markets are universities such as UCLA and Purdue, government research labs such as Los Alamos, Lawrence Livermore and White Sands, and private companies such as Hughes, Lockheed Martin, Boeing, and Arco.
As the demand continues for high performance computing (HPC) those skilled in the art have recognized a significant need for a thin server solution designed to accommodate very large network applications. The present invention fulfills this need.
SUMMARY OF THE INVENTION
An expandable multiprocessor computer system is provided incorporating a master computer, control computers and servant computers housed in modular units. Servant computers of the super scalable multiprocessor system act as cells of a singular entity enhancing system performance over conventional individually networked computers. Application information is transported between servant computers by way of a standard ultra high speed primary network as opposed to a specialized ultra high speed back plane. The design of the primary network allows for parallel internal data transfer in single modules, and allows for parallel internal data transfer between multiple modules through the provision of control computers in each module. Administration information is transported between the master computer, control computers and servant computers by way of a secondary network.
In a presently preferred embodiment, the inventive modular parallel processing system comprises, in combination:
a) a master computer;
b) a plurality of control computers each having means to route information between system modules;
c) a plurality of single board servant computers each having means to execute processing in accordance with a start instruction to retrieve identification information and to load a prescribed operating system;
d) primary network means for transporting application information between the servant computers by standard ultra high speed network, the primary network means being capable of parallel internal data transfer within single modules and parallel internal data transfer between multiple modules and
e) secondary network means for transporting administrative information between the master computer, control computers and servant computers to control the distribution of IP addresses when the system is initiated to facilitate identification of servant computers and to enable remote booting of such servant computers.
In accordance with the present invention, a unique expandable multiprocessor system is provided which enables ISPs to cost-efficiently host typical end-user services—such as Web sites, email, calendaring and scheduling—and scale those services quickly and easily to meet surges in demand. The novel system also provides an opportunity for ISPs to take advantage of business application outsourcing. For example, Enterprise Resource Planning (ERP), sales force automation, and call center applications are typically extremely large and complex, and are not easily condensed or simplified for customers who do not require the full functionality of the application. ISPs can use the inventive systems to host these applications and offer functionality as needed to their customers. This strategy enables smaller companies to take advantage of enterprise-class applications and services.
REFERENCES:
patent: 5613070 (1997-03-01), Born
patent: 6101419 (2000-08-01), Kennedy et al.
patent: 6460128 (2002-10-01), Baxter et al.
Jeppesen Karsten
Smith Mitchell
Frisenda, Jr. Frank
Ray Gopal C.
Total Impact
LandOfFree
Super scalable multiprocessor computer system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Super scalable multiprocessor computer system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Super scalable multiprocessor computer system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3231170