HPC in the Land of 24/7

By Michael Feldman

November 23, 2007

More businesses than ever are employing high performance computing capabilities to fulfill their mission-critical needs. While many of these companies aren’t using traditional technical computing, they still require a level of processing power, networking performance or storage scale that necessitates HPC assets. In most cases, the systems are not being used to produce a single answer or model a specific problem, but rather provide a continuous high performance capability for processing real-time transactions. In this type of environment, pure performance is not enough; marrying HPC with mission-critical computing is the real challenge.

Examples of such businesses include Wal-Mart, NASDAQ, and FedEx, three companies that shared their experiences with high performance computing at a Masterworks session at SC07 in Reno last week. The session was organized with the help of the Council on Competitiveness, an NGO that focuses on U.S. economic competitiveness opportunities and challenges.

NASDAQ — Speed, Cost and Reliability are Key

As executive vice president of Operations and Technology and chief information officer of NASDAQ since 2005, Anna Ewing has witnessed a rapid transformation of financial market exchanges. Although the industry is now extremely high-tech, it’s been slow to become globalized in the manner of most other industries. Here in the U.S., and even more so, in other countries, the exchanges have been maintained and protected as near monopolies by their government benefactors. Today though, the globalization of market exchanges is occurring in parallel with the rapid increase in electronic trading volume. In this environment, transaction speed, data throughput and low latency messaging are the technological features that give exchanges their competitive edge.

The most immediate challenge for NASDAQ is to keep up with the message data as electronic exchange traffic continues to skyrocket. Ewing says the exchange use to double its data traffic every year; now it’s every six months. The interconnectedness of the global markets is also stressing the system. Thanks to the near instantaneous transfer of market data, disruptive financial events quickly ripple through the world’s markets. In this volatile environment, predictability becomes a real asset and users gravitate to those exchanges where they know the trades can be executed reliably.

According to Ewing , their target is Four Nines (99.99 percent) reliability and they’ve been tracking to Five Nines (99.999 percent). Immediately after 9/11, the NASDAQ systems remained operational, thanks to a virtualized model and computing resources that were distributed across the country. But a lot of their customers were not nearly so fortunate, either because they relied on New York assets or because the redundant systems they had in place had never been tested, and didn’t perform as expected. Because of this and the general chaos of the financial environment, NASDAQ ended up voluntarily shutting down the exchange after 9/11. The lesson for NASDAQ was to include their customers in their business continuity planning and testing.

Because of the ubiquity of Internet applications and recent changes to the market regulatory framework, the barriers to automated trading have lowered dramatically. Achieving low latency market data messaging has becomes a critical feature for attracting traders. At NASDAQ, they’re constantly looking at ways for improving the messaging infrastructure to shave time off transactions. Ewing says they now can provide less than a 1 ms round-trip per message. In an effort to shave microseconds of latency from trades, some customers are collocating in NASDAQ facilities to get an edge over their competitors coming through the WAN.

“From a technology perspective, speed, reliability and low-cost are the life blood of our market,” says Ewing ” On any given day, we will process over two billion transactions at sub-millisecond speeds, at rates of over 200,000 transactions per second.”

Because of the rapidly increasing volumes of transactions, scaling their computing infrastructure becomes a continuous process, not something to be addressed every three or four years as equipment becomes obsolete. NASDAQ relies almost exclusively on commodity platforms, along with their own customized software. Using this model, over the last several years they’ve been able to reduce their cost base by 70 percent.

“There’s nothing fancy about our platforms,” explains Ewing. “It’s the software and network engineering that we perform that is, quite frankly, our core competence — our secret sauce, if you will.”

Wal-Mart — The Challenge of the 410 Billion Row Table

Nancy Stewart, senior vice president and chief technology officer of Wal-Mart Stores Inc., is in charge of the company’s infrastructure, operations and technology roadmap. That turns out to be quite a responsibility. Wal-Mart is the largest retailer in the world, a $370 billion company, whose revenue is larger than IBM, Intel, Microsoft, HP and Dell combined. The company is on track to become the first $1 trillion dollar company within the next few years.

Although Wal-Mart does not talk specifics about the scope of the computing and storage infrastructure it administers, in order to manage their inventory and supply chain, the company must process a 410 billion row table to figure out what is going to end up on its world-wide store shelves on any given day. The data has to be massaged very quickly, so that inventory control can react to real time events, like disasters, man-made supply disruptions or seasonal demand spikes. While the stores themselves may close, the company’s IT infrastructure is up 24/7.

“The value for us in using high performance computing is related to the fact that we have one of the largest data stores in the world,” says Stewart. “In terms of using that data store, in any given two hour period we have to process over two petabytes of data.”

Wal-Mart develops about 80 percent of their software in-house to maintain the level of reliability and availability that they require. When your company is netting $2 billion per hour on the day after Thanksgiving, downtime is not really an option. To work with Wal-Mart, suppliers and other partners have to match the retailer’s devotion to continuous availability. Because of the magnitude of transactions and the cash flow, Wal-Mart doesn’t maintain service level agreements (SLAs) with their computing partners. According to Stewart, none of them could afford the penalties involved with any downtime.

The ongoing problem for Wal-Mart is that their inventory management database has become so large that they’ve maxed out on their ability to handle it. The company’s application represents the “Grand Challenge” of real-time transaction processing. A trillion-row table, which they foresee in the next few years, is going to be difficult to process in real time. What they’re really looking for are predictable tools that can scale to their future needs. In truth, Stewart would prefer even faster turnaround on the inventory they currently manage.

“I really need to be able to mine the data much more quickly than I am now” admits Stewart…. “I’m not getting that today.”

FedEx — Logistics Planning on a Grand Scale

Kevin Humphries is the senior vice president of Technology Systems for FedEx Corporate Services and is responsible for setting technology direction as well as providing data center, network and field infrastructure support. The company’s computing technology orchestrates the delivery of millions of items each day around the world, using a fleet of over 600 aircraft and 75,000 motorized vehicles.

According to Humphries, the only way they’re able pull off this global logistics puzzle is to employ HPC simulation and modeling to help plan the FedEx routes. Trucks and planes have to be continually shuffled from place to place in the most efficient manner possible to make timely deliveries and to optimize resources. It’s not just a mega-version of the traveling salesman problem. In addition to the complex routing, the company has to deal with unforeseen events like weather and equipment breakdowns. On top of that, FedEx has essentially no control over shipping demand at any given time. But it’s the scope of the problem that precipitates the need for HPC.

“We have to take everything that comes our way,” says Humphries. “That creates about 30 million origin-destination pairs that have to be planned 24/7 every hour of the day, over all the assets that we own.”

The initial logistics plan for using the assets is performed with traditional HPC cluster tools well in advance of the actual shipments. As the time winds down to the day of execution, the model is continuously refined (some on grid platforms) to support a real time response. The refined model has to react to environmental conditions, like weather, mechanical breakdowns and infrastructure problems. An extremely high capacity computing environment is used to coalesce all the information in real time.

Humphries main frustration with high performance computing technology is its uniqueness. Businesses like FedEx would like to see their HPC assets seamlessly embedded into their overall enterprise infrastructure rather than have to be treated as an island of resources devoted to solving specialized problems. He thinks that transition is occurring, but they still struggle with some of the distinctive aspects of HPC, especially as it pertains to their cluster computing resources. The mainframes of the past were much easier to deal with compared to a system with thousands of nodes, where the job has to split up into little pieces. Further constraining the use of these systems is the limited pool of talent that can manage those resources.

“I don’t know where that changes though,” says Humphries. “It’s not something that every kid is going to learn in college and it’s not something everybody is going to learn on the job.”

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!

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…

Nvidia Appoints Andy Grant as EMEA Director of Supercomputing, Higher Education, and AI

March 22, 2024

Nvidia recently appointed Andy Grant as Director, Supercomputing, Higher Education, and AI for Europe, the Middle East, and Africa (EMEA). With over 25 years of high-performance computing (HPC) experience, Grant brings a 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…

Houston We Have a Solution: Addressing the HPC and Tech Talent Gap

March 15, 2024

Generations of Houstonian teachers, counselors, and parents have either worked in the aerospace industry or know people who do - the prospect of entering the fi 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