HPC in the Cloud Offers Flexible, Scalable Solutions for Enterprises and Research Institutions

Print Friendly, PDF & Email

In this guest article, our friends at Intel discuss how HPC as a Service (HPCaaS) reduces barriers to HPC and offers a cost-effective extension to on-premise solutions.

High performance data analytics (HPDA), simulation and modeling, and artificial intelligence (AI) enable significant benefits like quicker business insights, the ability to design and bring products to market faster, or real-time monitoring of financial transactions for fraud. In the past, on-premise HPC systems served as the sole mechanism to run demanding workloads like these. Today, though, moving HPC workloads to the cloud is gathering momentum as an alternate option. Cloud service providers now offer customers viable options with the scalability and speed necessary for HPC workloads.

As HPC-based usage scenarios increase, organizations of all sizes look at how cloud fits with their workload needs, expertise, and budget. For those institutions just beginning the journey toward HPC, cloud-hosted HPC solutions can offer a stellar starting point. Those institutions with detailed experience deploying and managing large-scale HPC systems on-premise, also benefit with capabilities to supplement their current HPC infrastructure capacity on-demand.

For those considering the cloud as an option to further their business or research endeavors with HPC, this article offers additional perspectives to help find a solution that best meets your organization’s needs.

Tapping into HPC as a Service

The cloud offers paths for HPCaaS to access performant, scalable, and even turnkey HPC solutions for demanding workloads. Cloud service providers like Amazon Web Services (AWS), the Google Cloud Platform, Microsoft Azure, and others provide the latest Intel HPC hardware and software performance libraries. In turn, this means their HPCaaS customers can access the most advanced HPC technologies now as an alternative to or augmentation of capital expenditure for new on-premise hardware and software.

Azure, for example, offers HPC solutions that serve scientific audiences and enterprises in vertical industry segments like aerospace, energy, financial services, insurance, manufacturing, and more. Microsoft Azure’s latest HC-series virtual machines (VMs) speed workloads with 44 Intel® Xeon® Platinum 8168 processor cores, using 8 GB of RAM per CPU core, underlying them [i]. H-series VMs prove especially helpful for complex computing scenarios benefitting from a performant message passing interface (MPI).

The combination of these elements offers users a highly-effective solution for intense workloads like computational fluid dynamics (CFD). Azure VMs also provide an excellent option for customer scenarios involving tasks like predictive analysis capability using deep learning or machine learning, accelerated rendering in engineering or media production, or visualization.

Specialized HPCaaS solutions

HPC in the cloud capability does not reside only in the realm of the world’s largest service providers, however. Specialized hosting companies like Advania recognize the need for HPCaaS and offer flexible options for their client base, too.

Based in Iceland, Advania specializes in HPCaaS offerings and is the first cloud service provider to offer instances based upon the 2nd Generation Intel Xeon Platinum 9200 processor. Their customers rely on HPCaaS to address their needs for data-intense workloads like simulation, computer-aided engineering, fluid dynamics, and more.

By differentiating themselves from the largest HPCaaS providers, Advania offers its customers unique benefits. First off, customers have the option of choosing a fully isolated, dedicated system. They have complete control of their virtual cluster down to the “bare metal” level, allowing them to assemble highly-customized clusters for unique workloads. Advania also enables end-to-end data encryption, ensuring that customer data remains private. This feature proves especially useful for those users tapping HPCaaS for proprietary research and development.

Advania’s solutions also offer their customers advantageous performance characteristics. This configurational approach, combined with the fast connections enabled by Intel® Omni-Path Architecture, helps accelerate customer workloads.

On top of these benefits, the combination of cool Iceland weather, readily available geothermal and hydroelectric energy, and Advania’s server optimizations gives their data centers the ability to run on one hundred percent renewable energy.

Running the right workload in the right place

A combination of an on-premise cluster and HPCaaS offers organizations greater flexibility for managing capacity for mission-critical workloads. Many elements weigh into a decision of the ideal hosting environment, including factors like peak usage times, regulatory and compliance issues, or changes in business approaches.

Organizations which have made investments in on-premise HPC systems control every aspect of the cluster. They can configure and optimize it to get the fastest results for their business-centric applications or research needs. While on-premise solutions remain ideal for some scenarios, an increasing number of organizations find HPCaaS’s elastic model a cost-efficient approach to extend their existing HPC system. For example, if users can “rent” HPCaaS resources needed to complete a memory-constrained job, the “hybrid” HPC method reduces the need for additional onsite hardware investments or developing workarounds for resource-constrained applications. HPCaaS also serves some organizations as a “sandbox” HPC environment that isolates a “test” project from production systems. The cloud provides options to control and prioritize workloads on local systems and the ability to handle peak demand that exceeds local capacity.

Underlying Intel Architecture

Organizations’ needs for on-premise or cloud-based HPC systems vary. However, 2nd Generation Intel Xeon Scalable processors, Intel® Optane™ DC Persistent Memory, and other ingredients offer the reliability, flexibility, and performance to get the job done.

For organizations considering on-premise HPC solutions or HPC in the cloud, please visit Intel.com/hpc for more information or check out this Gompute case study illustrating the benefits HPCaaS can offer.

For more complete information about performance and benchmark results, visit www.intel.com/benchmarks.

Intel technologies’ features and benefits depend on system configuration and may require enabled hardware, software or service activation. Performance varies depending on system configuration. Check with your system manufacturer or retailer or learn more at intel.com/hpc.

Intel, the Intel logo, and Xeon are trademarks of Intel Corporation or its subsidiaries in the U.S. and/or other countries.

*Other names and brands may be claimed as the property of others.

© Intel Corporation.

[i] https://docs.microsoft.com/en-us/azure/virtual-machines/linux/sizes-hpc#hc-series