ESnet at 30: Evolving Toward Exascale and Raising Expectations

By Tiffany Trader

December 10, 2015

In tandem with high-performance computing, high-speed nationwide and global infrastructure networks provide the essential backbone for today’s collaborative science workflows. In the United States, the Energy Sciences Network (ESnet) is the mission network of the U.S. Department of Energy. This high-performance, unclassified network that is managed by Lawrence Berkeley National Laboratory is moving into the newly-constructed Wang Hall on the Berkeley Lab campus.

ESnet links 40 DOE sites across the country and scientists at universities and other research institutions via a 100 gigabits-per second backbone network. One of these sites, the National Energy Research Scientific Computing Center (NERSC) has made the move to the Berkeley campus from its previous 15-year home in Oakland, California. ESnet has built a 400 gigabit-per-second (Gbps) super-channel between the Berkeley and Oakland sites to support this transition over the next year. This is the first-ever 400G production link to be deployed by a national research and education network, and will also be part of a research testbed for assessing new tools and technologies that are necessary to support massive data growth as supercomputers approach the exascale era.

“We are often the first organization to adopt new networking technologies because our scientists are really pushing the envelope when it comes to data transfer and access of large data sets,” commented ESnet Director Greg Bell in an interview with HPCwire. Along with Internet2, ESnet also built the first nationwide continental scale 100 Gbps network in 2012. “If you think of ESnet as being the national labs network, you can think of Internet2 as being the university network in the US,” Bell clarified.

In his role as ESnet director, Bell oversees all of the operational activities of one of the largest and fastest networks in the world — there are network engineers on call 24-7, a cybersecurity team, storage experts, data collection and data analysis activities, and efforts engaged in building out the network. Bell also oversees teams who help make the network useful to scientists. There is a team of people who build software tools to help the network be less of a black box. Then there is another team focused just on science engagement, helping scientists make the best possible use of the network and raising expectations about the network capabilities.

“This team directly engages with scientific collaborations large and small, but mostly large to medium-sized,” noted Bell, “and it also teaches scientists and networkers around the country and around the world best practices for networking so we can all build networks that are better and make it easier to move data and make it easier to accelerate scientific outcomes.”

Over the last ten years, the ESnet team has seen a move away from the sneakernet model, in which data is moved using a storage medium that is carried on a person or sent via a postal service.

“We aren’t ideologically opposed to sneakernet,” said Bell. “If you just need to move data once and you know you never need to access it again, it can sometimes be the most efficient solution, but in general, people need to move data over and over again, and they need to combine it with other data sets and they need to share it and they need to access it later and for that, networks are just great.

“We are trying to raise everyone’s expectations and let them know that networks can do much more than they could just a few years ago. In fact, the great vision that we have for networks is not only as a scientific instrument in their own right, but that they can glue together big scientific instruments like a particle accelerator or a light source and a computational facility, for example, a DOE supercomputer center. This enables a scenario where we can take data in real-time from the source and move it at high-speed over the network and process it in real-time at the supercomputer center so the scientists can get immediate feedback about the experimental parameters that they have chosen and then adjust them in real-time.”

“Doing this requires that the network glue together two or three other instruments,” Bell added. “If we can do that, we can make the DOE science complex and the US science complex more than the sum of its parts. We can enable discovery workflows that wouldn’t have been possible without excellent high-speed networks.”

ESnet then and now

ESnet will be 30 years old in 2016, which makes it one of the oldest networks in the world. “It actually predates the creation of the commercial network,” Bell shared. The DOE network was created at at time when two DOE science activities, one in high-energy physics and one in fusion energy — each had their own network before the Internet had really settled down into one technical architecture. In 1986, it was decided to create a single unified mission network and to chose a single architecture, which was TCP/IP, which is the way that the Internet evolved.

“ESnet was created out of the merger of these two domain specific science networks and since then, fusion and most especially high-energy physics has pushed us to be at the bleeding edge of networking for those 30 years,” Bell added.

The Department of Energy’s Office of Science funds nearly half of the physical science research in the US and provides about a billion dollars a year to university campuses. ESnet provides the high-bandwidth, reliable connections that link scientists at national laboratories, universities and other research institutions, enabling them to collaborate on some of the world’s most important scientific challenges within energy, climate science, and the origins of the universe.

The fundamental challenge for ESnet is keeping up with data growth, which has increased at a fairly steady exponential rate since its inception. Since 1990, ESnet’s average traffic has grown by a factor of 10 every 47 months, roughly along a Moore’s law growth curve. Last month, the network moved 36 petabytes of traffic.

What’s interesting, though, is that there has been a change in the source of this data as Bell explained. “It used to come from very large experiments like Large Hadron Collider (LHC) ATLAS and CMS detectors,” he said. “Now, it’s still coming from those large experiments but increasingly it’s coming from a lot more sources that are smaller and cheaper, for instance the detectors at the DOE Advanced Light Source beamline. Those are conceptually like the cameras in a mobile phone and they are getting much more high-resolution and the refresh rate is getting faster and faster. That compounded effect of high-resolution and faster refresh rates means that individual detectors are capable of sending out 10 Gbps or much more and soon this will be 80-100 Gbps.”

“So it actually is a tremendous challenge to engineer the network so it can grow cost-effectively,” said Bell. “We don’t have exponential budgets, we actually have at best linear budgets and sometimes flat budgets, so the question is how can we keep up with the demand.

“The light sources are just one example,” Bell added. “Tiny inexpensive genomic sequencers are producing a lot of data, as are environmental sensors, telescopes, and cosmology experiments, so for us it adds up to this exponential growth curve that is the fundamental challenge of ESnet, which is to evolve its architecture to accommodate and stay ahead of this growth curve.”

Over the years, ESnet has continued to rise to the challenge of supporting this exponential growth. ESnet5, the moniker for the current ESnet instantiation, is the 100 Gbps transcontinental and transatlantic network that was constructed a few years ago. They are now planning for the next network, ESnet6, which will probably need to use a different technology, according to Bell. To that end, he and his staff are keeping a close eye on developments in software-defined networking to produce more efficient use of the network as well as a consolidation of networking layers.

15-CS-1035 ESnet EuropeUS Map_v2_wkey

“Typically we used separate kinds of components at the optical layer and one at the higher layers, layers 2 and 3 that do switching and routing,” said Bell. “One path of innovation that we are watching closely is trends to consolidate those layers so you would buy one device instead of two that could handle the optical networking and also the traditional IP networking on top of that. It will probably be some combination of those two major trends that will produce the architecture that we will procure for the new network, but since it is still early in the process, this is still very conceptual.”

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire