<< Back to March 2004 CRN Table of Contents

[Published originally in the March 2004 edition of Computing Research News, Vol. 16/No. 2.]

By Horst Simon

BACKGROUND

Lawrence Berkeley National Laboratory has been a leader in scientific research in the United States for more than 70 years, winning nine Nobel Prizes and drawing leading scientists and technical staff from around the world. Today, the Lab's Computing Sciences organization is one of the premier groups of its kind in the country, with state-of-the-art infrastructure and expertise available for unclassified research projects that advance scientific discovery in all disciplines.

Situated in the hills above the University of California, Berkeley campus, with far-reaching views of the Bay area, the Lab is an inspiring setting for leading-edge scientific research. Ernest Orlando Lawrence, winner of the 1939 Nobel Prize in Physics for his invention of the cyclotron, founded the Lab in 1931 and promptly advanced the idea of doing research with multidisciplinary teams of scientists and engineers, treating all members as equal and valuable partners. This "big science" approach to solving problems is still alive and well at the Lab, and aptly describes its mission across all types of research. Likewise, Berkeley Lab has been a partner in many technological innovations in computing and networking.

The Lab's Computing Sciences organization is a key contributor to scientific research funded primarily by DOE's Office of Science and conducted at national labs and universities across the country, providing the computing and networking infrastructure essential to collaborative science. Comprising three divisions, Computing Sciences supports fundamental research in networking and computer science, and applied mathematics, and provides high-performance computational and networking tools to researchers in environmental and atmospheric research, fusion energy, structural biology, genomics, chemical and materials sciences, and high energy, nuclear, and plasma physics. In all of these fields, high performance computing (HPC) allows researchers to organize and analyze massive amounts of data in short periods of time, and design complex experiments to test theories that advance scientific understanding.

COMPUTING SCIENCES AT BERKELEY LAB

In order to accommodate the demands of advanced scientific research, the Berkeley Lab's Computing Sciences organization has specialized divisions to manage and allocate the computing tools and technologies to the DOE research community.

The National Energy Research Scientific Computing Center (NERSC) provides high performance computing, networking resources and expertise to DOE researchers nationwide. As the DOE's flagship facility for unclassified computing, NERSC provides leading-edge platforms and services, including the 6,656-processor IBM supercomputer "Seaborg" with a peak performance of 10 teraflop/s. NERSC also operates a High Performance Storage System with a total data archiving capacity of 8 petabytes. Beyond sheer computing power, NERSC's staff provides a full range of services, including scientific user support, 24x7 operation and visualization techniques to help users get the best performance and results from their applications. NERSC is currently used by some 2,300 researchers at national labs and universities scattered around the country.

The Computational Research Division (CRD) conducts applied research and development in computer science, computational science, and applied mathematics to create computational tools and techniques that support scientific research. Developments in all of these areas help improve the performance and efficiency of high-performance computing systems and the applications that run on them. The CRD staff collaborates directly with scientists-in fields ranging from materials sciences to climate modeling to astrophysics-to solve computational and data management problems. They also create visualizations to help scientists gain new physical insights and make the data more comprehensible. The DOE Science Grid is another fundamental part of CRD, and provides an advanced distributed computing infrastructure based on Grid middleware, and tools to enable the degree of scalability in scientific computing necessary for DOE to accomplish its missions in science.

The Information Technologies and Services Division (ITSD) supports computational science by managing the Lab's computing infrastructure, supplying networking and telecommunication services, and providing technical assistance to Lab scientists. ITSD also handles cyber security for the Lab, and has developed some of the most sophisticated monitoring and detection tools available and shares its expertise with other DOE labs and networks. ITSD also manages DOE's Energy Sciences Network (ESnet)-a high-speed, general-purpose IP-based production network that connects national laboratories, universities, and other institutions around the world-to support DOE research missions. ESnet services allow scientists to make effective use of unique DOE research facilities and computing resources, independent of time and geographic location.

LAB STAFF

In order to achieve these goals, the Computing Sciences organization brings together specialists from a variety of fields to collaborate in building customized applications and the expertise to make them run quickly and efficiently on leading-edge systems. Many of the Lab's researchers have a background in math, physics, chemistry and other traditional sciences, in addition to computer science, and the combination of these skills helps them design applications and tools to solve even the most complicated scientific puzzles.

The staff also brings together industry, academic, and lab experience, which makes for a valuable resource in designing and executing research projects. There are approximately 500 joint staff appointments with the University of California, Berkeley, ensuring that the Lab benefits from a steady flow of information between the two institutions. Given the dynamic nature of the work at the Lab, researchers and scientists find new areas to apply their skills and new projects where their expertise is needed. It is not unusual to find staff members who have been around for more than a decade, or two or three, and others who return to work not long after their retirement parties.

The Lab, in total, employs about 4,300 people, including nearly 1,000 staff scientists, 1,000 undergraduate students, graduate students, and postdoctoral fellows, and more than 1,500 technical and support personnel. In addition, each year the Lab hosts more than 2,000 participating guests.

ACHIEVEMENTS

From the earliest days of scientific computing, Berkeley Lab has taken a leadership role in deploying and applying systems for research. In the 1950s, Luis W. Alvarez opened a new era in high-energy physics research with his proposal to build a pressurized chamber filled with liquid hydrogen. Known as a "bubble chamber," this device would allow scientists to discover and study new particles. In his 1955 prospectus for such a facility, Alvarez became one of the first scientists to propose using computing devices for analyzing experimental data, even before such computers were actually available.

By the 1960s, Alvarez's vision was reality as LBNL researchers used computers to track some 1.5 million particle physics events annually and developed scientific computing techniques that were adopted by researchers around the world. This effort led to Alvarez receiving the Nobel Prize for Physics in 1968.

Here are some of the Lab's other computing and networking achievements:

• In 1974, LBNL connects its CDC 6600 computer to ARPANET, making it the first online supercomputer.
• In 1985, ARPANET (the forerunner of the Internet), was collapsing due to congestion of data transmission. The TCP congestion control algorithms from Van Jacobson's 1988 paper on congestion avoidance and control become Internet standards in 1989. In 1986, the network was plagued with routing instability between the many connecting systems, resulting in many transmissions being lost. Lab experts developed a tool to trace data packets along the way, allowing routing problems to be pinpointed and corrected.
• Van Jacobson and Steven McCanne of LBNL win one of R&D Magazine's 1995 R&D 100 Awards for development of a software toolpack that enables multiparty audio and visual conferencing via the MBone (Multicast Backbone).
• In 1998 Vern Paxson of LBNL is honored at a security conference for his paper "Bro: A System for Detecting Network Intruders in Real-Time." Two years later, logs generated by Bro help the FBI convict a hacker for breaking into DOE and DoD computers.
• In 1999, the first IPv6 address is assigned to ESnet, which plans to demonstrate the viability of using IPv6 to run scientific applications on the Internet.
• When DOE launches its Scientific Discovery through Advanced Computing (SciDAC) program in 2001, LBNL is selected as the lead lab for six projects. These include the Scientific Data Management Integrated Software Infrastructure Center, which will provide a coordinated framework for the unification, development, deployment, and reuse of scientific data management software; High-End Computer System Performance, which focuses on how specific scientific applications can best be run on high-performance computers; and a multi-laboratory collaborative project to develop, evaluate and deploy the needed services to support the DOE Science Grid.
• LBNL and UC Berkeley launch the Berkeley UPC compiler project in 2002 to develop a portable, high performance implementation of UPC for large-scale multiprocessors, PC clusters, and clusters of shared memory multiprocessors.
• In December 2003, NetworkWeekFusion names LBNL's Michael Bennett as one of the 50 most influential people in networking on a list dominated by leaders of network and computer companies.

GUIDANCE

Berkeley Lab is managed by the University of California for DOE. The Lab's funding is primarily from DOE and supports research in the Office of Science's six major programs: Advanced Scientific Computing Research; Basic Energy Sciences; Biological and Environmental Research; Fusion Energy Sciences; High Energy Physics; and Nuclear Energy Physics.

Horst D. Simon is Director, NERSC Center and Computational Research Divisions, Lawrence Berkeley National Laboratory.