Can HPE GreenLake for HPC Deliver a Simpler User Experience than Public Cloud?

Print Friendly, PDF & Email

Sponsored Post

A big part of public cloud’s appeal is the promise of a simple user experience, where HPC workloads are easy to deploy and teams can spend less time solving technical problems.

The cost of that simplicity may be performance. Recent HPE tests showed how Intel®-based AWS and Oracle cloud HPC solutions delivered less than half the performance per dollar of an HPE GreenLake for HPC solution powered by AMD EPYC™ processors, in High Performance Linpack benchmarks. For some workloads, that’s ok.

But does public cloud truly deliver a simpler user experience than a modern, dedicated HPC solution, such as HPE GreenLake for HPC?

Many people associate public cloud with simplicity. After all, there is no need to design, configure, or install your own system. You just consume it on demand. Yet today, AWS and its rivals are no longer the only places you can consume HPC this way. HPE GreenLake for HPC1 also provides pay-per-use HPC solutions optimized for your workloads—and they are deployed for you wherever you want them, with the record-breaking performance of AMD EPYC processors2.

So, which is really the fastest and simplest route to value for HPC teams who want to focus on solving problems, rather than technical hurdles? A recent HPE study compared the user experience when deploying HPC workloads on AWS, Oracle Cloud, and an HPE GreenLake for HPC solution. This post shares some of the key findings.

AWS and Oracle user experiences

The study team found that while public cloud HPC platforms may be easy to access and provision, it isn’t necessarily easy to get your projects up and running on them, or to optimize them.

On Oracle Cloud, the team found that compute instances needed to use Oracle Linux to run out of the box. The block storage service was found to have synchronization issues, making it unsuitable for IO-intensive workloads. And the team could not easily stop and start instances to maximize cost-efficiency. Billing continued when instance pools attached to NVMe storage instances were stopped, because it was not possible to stop NVMe storage without deleting the compute node.

The team found further issues with AWS. There was no way to efficiently use OpenMPI, an open source Message Passing Interface that is used by academics, researchers, and other industries. Amazon Elastic Fabric Adapter (EFA), an interface for HPC clusters, suffered from abnormally high latencies with OpenMPI during the study. And no direct support was available to help resolve these issues.

Hidden challenges of public cloud

These findings exemplify bumps in the road that may be common to many public cloud HPC experiences.

    • Infrastructure is less customizable than your own systems. So it can be difficult to add in features and use libraries that aren’t already supported.
    • Workloads still need configuration. Choosing an HPC-focused cloud compute instance is not the end of your setup work.
    • You’ll generally receive minimal support as standard. Cloud platforms aren’t here to partner you all the way to success—you take these services as they come.
    • It can feel like you’re losing control of your data. As your datasets grow, so do public cloud data egress charges. This can limit your freedom to move data wherever and whenever you want to.

These problems aren’t always visible until you try to get your project up and running, but they’re hard to ignore if they slow down your progress or limit your options.

What is HPE GreenLake for HPC?

In comparison, HPE GreenLake for HPC is designed to help you get the benefits of HPC without the deployment challenges. It’s a consumption-based solution that is fully managed and operated for you, just like public cloud. But you can also choose from the world’s most-proven, market-leading HPC systems, pre-integrated and optimized by HPE and AMD for your modelling and simulation workloads. And you can locate your infrastructure wherever it runs best whether on-premises, at your local data or co-location center.

It’s simplified in all the right places. Deployed and managed by HPE. Pre-configured and proven to run your workloads. And flexible enough to solve technical, financial, and compliance challenges.

It’s not easy to simplify the hybrid HPC as-a-service experience. HPE worked closely with partners including Plexus, a leader in container orchestration and cluster management in multi-cloud HPC environment, for over 5 years to achieve this. Plexus provides exceptional ease of use and flexibility in managing and operating hybrid and multi-cloud HPC environments.

HPE GreenLake for HPC solutions also deliver industry-leading performance and value with the power of AMD EPYC processors. AMD delivers the world’s highest performing x86 server CPUs3, which provide 60% higher core density and 2x the IO performance4, and faster time to value.

The obvious answer isn’t always the simplest

When looking for a simple HPC user experience, many of us would naturally think of public cloud. But on-premises solutions like HPE GreenLake for HPC, powered by AMD EPYC processors, have picked up the advantages of public cloud without compromising performance.

If you’d like to learn more, click below or talk to your HPE representative.


  3. AMD EPYC™ Processor World Records | AMD
  4. Comparison between highest performance EPYC (7763) and Ice Lake (Platinum 8380) processors: ,

About the Author

Max Alt, Distinguished Technologist and Director, Hybrid HPC, Hewlett Packard Enterprise

Max leads cloud-oriented HPCaaS initiatives including the GreenLake Cloud Services for HPC offering. Prior to joining HPE, Max was SVP AI & HPC Technology at Core Scientific that acquired Atrio in 2020 where Max was the CEO & Founder. Atrio created a leading-edge hybrid cloud platform for HPC orchestration, cluster, and container management.

Max has a unique background with almost 30 years of experience in software performance technologies and high performance computing. He is both an entrepreneur and a large-scale enterprise leader. Max founded several tech start-ups in the Bay area and he spent 18 years at Intel in various engineering and leadership roles including developing next generation super-computing technologies. Max’s strongest expertise are in computer and server architectures, cloud technologies, operating systems, compilers, and software engineering. Max received his BS in Math and Computer Science at Tel Aviv University and his master’s degree in Software Engineering from Carnegie Mellon University.