PROGRAM AREA OVERVIEW --
ADVANCED SCIENTIFIC COMPUTING RESEARCH

http://www.sc.doe.gov/production/octr/index.html

 

The Office of Advanced Scientific Computing Research (ASCR) supports research in computational technology and subprograms that underlie a variety of Department of Energy missions. 

ASCR's primary mission, carried out by the Mathematical, Information, and Computational Sciences subprogram, is to discover, develop, and deploy the computational and networking tools that enable researchers in the scientific disciplines to analyze, model, simulate, and predict complex phenomena important to the Department of Energy.  To accomplish this mission the program fosters and supports fundamental research in advanced scientific computing - applied mathematics, computer science, and networking - and operates supercomputer, networking, and related facilities.  The applied mathematics research efforts provide the fundamental mathematical methods to model complex physical and biological systems.  The computer science research efforts enable scientists to efficiently run these models on the highest performance computers available and to store, manage, analyze, and visualize the massive amounts of data that result.  The networking research provides the techniques to link the data producers; e.g., supercomputers and large experimental facilities with scientists who need access to the data.  The two topics that follow support this scientific computing mission.

Emerging science experiments in the DOE are expected to generate several petabytes of data, which will be transferred to geographically distributed terascale computing facilities for analysis and visualization by thousands of scientists across the world.  In addition, many emerging energy research problems require coordinated access to distributed resources – people, data, computers, and facilities.  Unlike today’s Internet that is optimized for low-speed web applications, this new emerging distributed terascale scientific computing environment calls for ultra-high-speed networks – networks that can deliver terabits/sec throughput to science applications.  This topic is focused on the development of ultra-high-speed transport protocols and network interfaces to support terabits networks and on technology to support coordinated resource sharing across those networks.  Grant applications must address scalability issues associated with proposed approaches by demonstrating how the resulting system will be operated from OC-48 to OC-192 (Optical Carrier levels 48 to 192).  Grant applications are sought only in the following subtopics:  

a.  Ultra High-Speed Transport Protocols—Grant applications are sought to develop radical new approaches to transport protocols – TCP (Transmission Control Protocol) and UDP (User Datagram Protocol) – to be used in overlay optical networks.  Such transport protocols, preferably all-optical, must be capable of delivering and sustaining multi-Gigabits/sec throughput to high-end scientific applications – in particular, high-speed data transfer applications. 

b.  Smart High-Speed Network Interface Cards—Grant applications are sought to develop intelligent, high-speed Network Interface Cards (NICs) to perform transport protocol off-loading at ultra-high-speed for emerging network technologies such as 10 GigE (Gigabit Ethernet) and all-optical lambda networks.  Adapters using this technique would improve host processor efficiency, increase data throughput, and reduce latency.  The emergence of fast, cheap, embedded processors present an opportunity for inexpensive processing to occur on the network interface.  

c.  High-Speed Intrusion Detection Systems—Grant applications are sought to develop high-speed, real-time intrusion detection systems capable of handling OC-48 and OC-192 traffic. These new intrusion detection systems should be automated with Artificial Intelligence (AI) techniques such as fuzzy logic, neural networks, and genetic algorithms to perform computer-assisted intrusion detection and corrective actions.

References:

1.       ESnet:  The Energy Sciences Network, U.S. DOE, Office of Science, http://www.es.net  

2.       Feng, W. and Tinnakornsrisuphap, P., “The Failure of TCP in High-Performance Computational Grids,” SC 2000:  High-Performance Networking and Computing Conference, November 2000.  (URL:  http://public.lanl.gov/radiant/website/pubs/traffic/sc00.pdf

3.       Fisk, M. and Feng, W., “Dynamic Right-Sizing:  TCP Flow-Control Adaptation,” poster at the 14th Annual ACM/IEEE SC2001 Conference, November 2001.  (URL:  http://public.lanl.gov/mfisk/papers/drs-sc01-poster.pdf)

4.       Floyd, S., HighSpeed TCP for Large Congestion Windows, Internet Engineering Task Force (IETF) Draft, August 2002.
(URL:  http://www.ietf.org/internet-drafts/draft-floyd-tcp-highspeed-01.txt 

5.       IEEE P802.3ae 10Gb/s Ethernet Task Force
Institute of Electrical and Electronics Engineers, Inc., March 2002 http://grouper.ieee.org/groups/802/3/ae/public/index.html

6.       IETF RFC2823:  PPP Over Simple Data Link (SDL) Using SONET/SDH with ATM-Like Framing, Internet Society, May 2000  http://www.faqs.org/rfcs/rfc2823.html

7.       National Coordination Office for Information Technology R&D, http://www.itrd.gov  

8.       Net100, http://www.net100.org  

9.       The WEB100 Project, http://www.web100.org  

10.   Tinnakornsrisuphap, P., et al., “On the Burstiness of the TCP Congestion-Control Mechanism in a Distributed Computing System,” 20th International Conference on Distributed Computing Systems (ICDCS'00), April 2000.  (URL:  http://public.lanl.gov/radiant/website/pubs/traffic/icdcs00-tcp.pdf)

11.   Office of Science, U.S. Department of Energy
http://www.science.doe.gov 

 

41. SCALABLE MIDDLEWARE AND GRID TECHNOLOGIES

Advances in high performance network capabilities and collaboration technologies are making it easier for large geographically dispersed teams to collaborate effectively.  This is especially important for research teams that use major computational resources, data resources, and experimental facilities supported by DOE.  The importance of collaboratories is expected to increase in the future.  However, significant research questions must be addressed if collaboratories are to achieve their potential, namely, by providing:  (1) remote access to facilities that produce petabytes/year; (2) remote users an with an experience that approaches "being there;" (3) remote visualization of terabyte to petabyte data sets from computational simulation; and (4) effective remote access to advanced scientific computers.  Research and software tool development are needed to support coordinated and dynamic resource sharing in areas such as resource discovery, resource access, authentication, authorization, accounting, etc. in the areas listed below.   Any tools or services developed should be interoperable according to emerging standards from the Global Grid Forum.  Grant applications are sought only in the following subtopics:  

a.  Scalable Middleware Technologies—Grant applications are sought to develop scalable middleware technologies that will (1) enable universal, ubiquitous, easy access to remote computing resources and scientific instruments; (2) facilitate collaboration among distributed science teams; and (3) enable a new generation of distributed high-end applications.  Areas of interest include, but are not limited to, secure directory services, scalable authentication/authorization services, deployable LAN and WAN QoS services, data streaming management, multicast and efficient broadcast capabilities, automatic resource discovery protocols, remote data access services, and network-attached memory and storage systems.  

b.  Scalable Grid Technologies—Grant applications are sought to develop scalable grid technologies to support the emerging distributed computing network that provides dependable, consistent, pervasive, scaleable, and efficient access to various resources integrated into a distributed infrastructure that can be accessed wherever and whenever by DOE scientists.  These resources include visualization systems, computer systems, data storage and archive systems, and scientific instruments.  Areas of interest include, but are not limited to, collaborative visualization systems, collaborative problem solving services, application level fast data transfer toolkits, real-time analysis, group collaboration, co-scheduling distributed resources, grid accounting and billing mechanisms, data management tools, science portals, on-line instrumentation, and fast data transfer management services.   

References:  

1.       DOE Science Grid, U.S. DOE Office of Science
http://doesciencegrid.org/

2.       Foster, I. and Kasselman, C., eds., The Grid:  Blueprint for a New Computing Infrastructure, San Francisco, CA:  Morrgan Kaufmann Publishers, 1999.  (ISBN: 1-55860-475-8) 

3.       Global Grid Forum (GFF) http://www.globalgridforum.com/ 

4.       Globus Project:  Related Papers http://www.globus.org/research/papers.html

5.       Particle Physics Data Grid, U.S. DOE Office of Science, http://www.ppdg.net  

6.       SciDAC (Scientific Discovery through Advanced Computing), U.S. DOE Office of Science
http://www.scidac.org  

7.      U.S. Department of Energy, Office of Science
http://www.science.doe.gov/