HPC as a Service: Lessons Learned

By Wolfgang Gentzsch and Burak Yenier

December 10, 2012

After a fast-paced three months, round 1 of the HPC Experiment (also known as the Uber-Cloud Experiment) concluded last month, with more than 160 participating organizations and individuals from 25 countries, working together in 25 international teams. In this article we present their main findings, challenges, and their lessons learned.

The aim of the Uber-Cloud Experiment is to explore the end-to-end process of accessing remote computing resources in HPC centers and in HPC clouds as well as to study and overcome the potential roadblocks.

The experiment kicked off in July 2012 and brought together four categories of participants: the industry end-users with their applications, the software providers, the computing and storage resource providers, and the experts. We set up an end-user project by first selecting an end-user and his software provider, assigning an HPC/CAE expert, and matching a suitable resource provider to complete the team. Each team’s goal was to complete the project, and to chart the way around the hurdles they identified.

End users can achieve many benefits by gaining access to additional compute resources beyond their current internal resources, such as workstations. Arguably the most important two are the benefits of agility gained by speeding up product design cycles through shorter simulation run times, and those gained by the superior quality achieved by simulating more sophisticated geometries or physics, or by running many more iterations to look for the best product design.

Tangible benefits like these make HPC and more specifically HPC-as-a-Service (HPCaaS) very attractive. But how far are we from an ideal HPCaaS or HPC in the cloud model?

Honestly, at this point, we don’t know. However, in the course of this experiment, following each team and monitoring its challenges and progress, we’ve collected some excellent insight into these roadblocks and how our 25 teams have tackled them.

The main approach for this experiment is to look at the end-users’ project and select the appropriate resources, software and expertise that match those requirements.

During the three months of the experiment, we were able to build 25 teams each with a project proposed by an end user. These teams were: Team Anchor Bolt, Team Resonance, Team Radiofrequency, Team Supersonic, Team Liquid-Gas, Team Wing-Flow, Team Ship-Hull, Team Cement-Flows, Team Sprinkler, Team Space Capsule, Team Car Acoustics, Team Dosimetry, Team Weathermen, Team Wind Turbine, Team Combustion, Team Blood Flow, Team Turbo-Machinery, Team Gas Bubbles, Team Side impact, Team ColombiaBio, and Team Cellphone.

The final report, available to all of our registered participants, contains the use cases of many of the teams offering valuable insight through their own words. We look forward to future rounds of the experiment where this accumulating knowledge will yield ever more successful projects.

We recognize that every end-user project requires a slightly different approach, a variety of software and compute resources, a certain expertise to lead the end-to-end process, and a tailored schedule. To be able to keep the entire experiment consistent we asked each team to follow a common roadmap. The expert assigned to each team is the guide in following this roadmap. It calls for communication with the organizers at certain points, although generally the teams are autonomous and make their own decisions.

Based on the roadmap we defined going into round 1 of the experiment, the teams followed six steps to reach their goal:

Step 1. Define the end-user project. The end-user together with the expert and software provider jointly defined the project. Based on this information, as organizers we assigned the appropriate resources to the project.

Step 2. Contact the resource provider and set up the project environment. The expert contacted the computing resource and performed activities such as assisting in software and license installations, creation of user accounts, and configuration of the project environment.

Step 3. Initiate the end-user project execution. The expert assisted the end-user with uploading the necessary data, code and configuration files to the remote resource(s). The expert then worked with the resource provider to queue the project up on the HPC system.

Step 4. Monitor the project. The expert remained engaged with the resource providers and at any time had up to date information about the status of the project.

Step 5. Review results with the end-user. The expert assisted the end-user in downloading the results from the resource provider’s environment and discussed the results with the end-user. If any rework or rerun was required it was completed by executing steps 2-5 again.

Step 6. Document findings. During the entire lifecycle of the project, there occurred hurdles, friction and failure points and the expert documented these findings.

Intentionally, we performed the first round of this experiment manually, that is, not via an automated service, because we believe the technology is not the challenge anymore; rather it’s the people and their processes, and that’s what we wanted to explore. We are continuously improving the roadmap to successful completion of our projects.

The teams reported the following main roadblocks and provided information on how they resolved them (or not):

  • Security and privacy, guarding the raw data, processing models and the results
  • Unpredictable costs can be a major problem in securing a budget for a given project
  • Lack of easy, intuitive self-service registration and administration
  • Incompatible software licensing models hinder adoption of Computing-as-a-Service
  • High expectations can lead to disappointing results
  • Lack of reliability and availability of resources can lead to long delays

Just like all other participants, we as the organizers, treated the experiment as a learning opportunity. In our report we have also summarized what we’ve found to be shortcomings of the experiment as we put it together in round 1. Learning from these shortcomings we have improved the experiment for round 2. To be specific, we discussed and provided solutions for the following shortcomings:

All participants are professionals with busy schedules and the experiment is not their primary job, so they could only dedicate a few hours per week to the experiment

  • Vacations delayed most of the teams’ progress, especially in the beginning (August) of the Experiment
  • Some resource providers ran into resource crunches which delayed team projects
  • Some of our projects ran into long delays since the project and the resource provider weren’t the best match possible
  • Some resource providers struggled with the installation of an application
  • Other resource providers had difficulties with providing network access through complex network connections
  • Resource providers differ in their service philosophies
  • Simply getting started was a challenge
  • A few teams struggled with figuring out which team member needs to do what and when
  • Team forming was one of the steps, which took the longest amount of time, each team member needed to exchange significant amounts of information about their background, capabilities, expectations, availability, and commitment levels with one another before the project could even kick off
  • Finally, manual processes are just slow; they consumed days, sometimes weeks especially because the various technology and people resources were inherently remote, each with different priorities

We hope that our participants will extract value out of the experiment and the final report. They certainly deserve to do so in return for their generous contributions, support and participation. We now look forward to round 2 of the experiment with its already over 250 participants and the learning that it will result in.

If you are interested in participating in round 2 or just want to monitor its progress, you can register at http://hpcexperiment.com.  You can also go there to get the final report for round 1, which details the results and recommendations.

About the Authors

Wolfgang Gentzsch and Burak Yenier are the creators and facilitators of the Uber-Cloud Experiment. Wolfgang is an HPC veteran. Having worked in leading positions in research, academia and industry for some 30 years, Wolfgang is now an HPC consultant and the chairman of the ISC Cloud conference series for HPC and Big Data in the Cloud. Burak is the vice president of operations at CashEdge, a software-as-a-service company in Silicon Valley, which provides innovative payments and aggregation solutions to financial institutions.


Related Articles

Half-Time in the Uber-Cloud

The Uber-Cloud Experiment

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