Pink Hat companions with US Division of Vitality Laboratories to bridge the hole between high-performance computing and cloud environments

Collaboration to assist set the stage for the arrival of exascale supercomputers by establishing finest practices for working next-generation HPC workloads

RALEIGH, NC, Might 31, 2022–(BUSINESS WIRE)–Pink Hat, Inc., the world’s main supplier of open supply options, right now introduced that it’s working with a number of U.S. Division of Vitality (DOE) labs to reinforce cloud native requirements and practices in high performance computing (HPC)together with Lawrence Berkeley Nationwide Laboratory, Lawrence Livermore Nationwide Laboratory and Sandia Nationwide Laboratories.

HPC adoption is increasing past conventional use circumstances. Advances in synthetic intelligence, machine studying, and deep studying, in addition to computational and data-driven analytics, are driving higher curiosity and wish for organizations to be in in a position to run scalable containerized workloads on conventional HPC infrastructure. Based on {industry} analyst agency Hyperion Analysis, roughly one-third of all HPC system income might be spent on AI-centric techniques by 2025, posting a five-year CAGR of almost 23%. 12 months.1, pushed by the inflow of AI workloads. Moreover, almost 20% of HPC customers’ HPC-enabled AI workloads are presently working within the cloud.2

Pink Hat is a pacesetter in cloud-native innovation hybrid and multicloud environments, whereas the labs perceive the distinctive wants and necessities of large-scale HPC deployments. By establishing a standard basis of know-how finest practices, this collaboration goals to make use of standardized container platforms to bridge HPC and cloud computing footprints, serving to to shut potential gaps in constructing cloud-enabled HPC purposes whereas creating widespread utilization patterns for {industry}, enterprise, and HPC deployments. .

Working with the labs, Pink Hat will deal with advancing 4 particular areas that tackle present gaps and assist lay the inspiration for exascale computing, together with standardization, scalability, cloud-native software improvement and container storage. Listed below are examples of collaborative initiatives between Pink Hat Labs and DOE:

Bringing normal container applied sciences to HPC
Pink Hat and Berkeley Lab’s Nationwide Vitality Analysis Scientific Computing Heart (NERSC) acknowledge the significance of standards-based options to allow computing innovation, particularly when applied sciences have to span from edge to cloud to HPC environments. From container safety to scaling containerized workloads, widespread and accepted practices assist HPC websites get essentially the most out of container applied sciences. To raised tackle the distinctive necessities of large-scale HPC techniques and pave the best way for organizations to have the ability to reap the benefits of containers in exascale computing, Pink Hat and NERSC collaborate to enhance Podmana daemonless container engine for creating, managing, and working container photos on a Linux system, to permit it to switch NERSC’s customized improvement runtime, Shifter.

Run Kubernetes at scale
Pink Hat has labored with Sandia Nationwide Laboratories on the Super Containers undertaking for a number of years, working to make Linux containers and different constructing blocks of cloud-native computing extra simply accessible to compute-intensive operations. On this prolonged collaboration, Pink Hat and Sandia Nationwide Laboratories intend to discover large-scale Kubernetes-based infrastructure deployment scenariosoffering easier, well-defined mechanisms for delivering containerized workloads to customers.

Bridging conventional HPC duties with cloud-native workloads
Pink Hat and Lawrence Livermore Nationwide Laboratory are collaborating to convey HPC job schedulers, comparable to Flowto Kubernetes by way of a standardized programming interface that helps IT groups supporting compute-intensive operations to higher handle conventional parallel workflows alongside containerized dutiestogether with how this mixture of applied sciences works with low-level {hardware} gadgets, comparable to accelerators or high-speed networks.

Reinventing container storage
For containers for use successfully on business HPC and cloud assets, a set of normal interfaces are wanted to deal with numerous container picture codecs and supply entry to distributed file techniques. Pink Hat and the three DOE Nationwide Laboratories goal to outline the mechanisms by which container photos will be migrated and deployed with different container engines, permitting customers to freely transfer their purposes to well-liked container runtime platformsin addition to creating mechanisms that enable containers to make use of distributed file techniques as persistent storage.

Because of the collaboration and Red Hat’s experience supporting some of the world’s most powerful supercomputersHPC websites will be capable of summary away the immense complexities that their environments can current, benefiting the vary of US exascale machines deployed by the DOE.

Supporting quotes
Chris Wright, senior vice chairman and chief know-how officer, Pink Hat
“The HPC group has served as a proving floor for compute-intensive purposes, early adopting containers to assist tackle a brand new set of scientific challenges and points. This has led to the shortage of standardization on completely different HPC websites, creating obstacles to constructing and deploying containerized purposes able to successfully spanning large-scale HPC, business, and cloud environments, whereas profiting from rising {hardware} accelerators. By means of our collaboration with main labs , we’re working to take away these obstacles, paving the best way for the discharge of next-generation HPC workloads.”

Earl Joseph, Ph.D., Managing Director, Hyperion Analysis
“Excessive-performance computing infrastructure should adapt to the calls for of right now’s heterogeneous workloads, together with workloads that use containers. Pink Hat’s partnership with DOE Labs is designed to allow the following technology of HPC purposes to run in exascale-scale containers whereas utilizing distributed file system storage. , offering a wonderful instance of collaboration between {industry} and analysis leaders. »

Shane Canon, Principal Engineer, Lawrence Berkeley Nationwide Laboratory
“Collaborating with the Podman group and Pink Hat engineers helps us discover and co-develop enhancements that may enable Podman to scale and work for the most important HPC workloads. We have already demonstrated this on 512 GPU nodes on Perlmutter. NERSC sees a convergence of HPC and cloud-native workloads, and Podman will be an essential device to assist bridge these two worlds.”

Bronis R. de Supinski, Chief Know-how Officer, Lawrence Livermore Nationwide Laboratory
“Excessive-performance computing infrastructure is diversifying and more and more getting used to run non-traditional HPC workflows. We have to present mechanisms for scheduling numerous sorts of workflows and we anticipate container orchestration frameworks comparable to Kubernetes and Pink Hat OpenShift to be an essential a part of the software program ecosystem successfully contributing to the convergence of HPC and cloud domains.”

Andrew J. Younge, Ph.D., R&D Supervisor and Laptop Scientist, Sandia Nationwide Laboratories
“Sandia and the DOE see an elevated have to help extra numerous HPC workloads past conventional batch modeling and simulation codes. This forces us to search out new and modern methods to allow providers, duties, and knowledge persistence fashions beneath shut coordination. with present simulation capabilities. Moreover, workload portability stays an essential consideration the place containers at the moment are a key a part of our code deployment technique. Sandia’s collaboration with Pink Hat on Podman and Kubernetes-based OpenShift permits us to discover approaches to offering modeling and simulation capabilities as a service to Sandia’s designer and analyst communities.”

Further Assets

Join with Pink Hat

About Pink Hat, Inc.
Red Hat is the world’s main supplier of enterprise open supply software program options, utilizing a community-based method to ship dependable and high-performance Linux, hybrid cloud, containers and Kubernetes applied sciences. Pink Hat helps clients combine new and present IT purposes, develop cloud-native purposes, standardize on our industry-leading working system, and automate, safe, and handle advanced environments. Prime help, coaching, and consulting providers make Pink Hat a Fortune 500 Trusted Advisor. As a strategic accomplice to cloud suppliers, system integrators, software distributors, clients, and open supply communities, Pink Hat may also help companies put together for the digital future.

Pink Hat Ahead-Wanting Statements
Apart from historic data and discussions contained on this press launch, statements contained on this press launch could represent forward-looking statements throughout the that means of the Personal Securities Litigation Reform Act of 1995. Ahead-looking statements are based mostly on the the corporate’s present assumptions concerning future enterprise and monetary efficiency. These statements contain quite a few dangers, uncertainties and different components that would trigger precise outcomes to vary materially. Any forward-looking assertion on this press launch speaks solely as of the date on which it’s made. Besides as required by legislation, the Firm undertakes no obligation to replace or revise any forward-looking statements.

Pink Hat, the Pink Hat brand, and OpenShift are logos or registered logos of Pink Hat, Inc. or its subsidiaries in the US and different international locations. Linux® is the registered trademark of Linus Torvalds in the US and different international locations.

1 Supply: Hyperion Analysis, “Global HPC-Based Artificial Intelligence (AI) MarketForecast, 2020-2025

2 Supply: Hyperion Analysis, “HPC and containers – an intriguing combination

See the supply model on


Daniel Thompson
RedHat, Inc.
+1 978-589-1059

Leave a Reply

Your email address will not be published.