Which model best describes the environment in which you want to run HPC workloads… on prem, hybrid cloud, HPC as a service, or a mix? Find out what they all mean and what version will best meet your need.
At HPE, we describe HPE GreenLake for HPC as “the cloud that comes to you,” and customers often have questions about this. Does HPE build its own cloud, or are we a cloud service provider? The difference sounds subtle, but is actually significant. Which cloud attributes are important to your organization, and are they delivered by HPE GreenLake for HPC? And, what does it mean to consume your HPC infrastructure as a service? You’ll find answers to these questions and others in this blog.
What is HPE GreenLake for HPC?
HPE GreenLake for HPC is a range of hybrid cloud solutions built on our industry leading HPC portfolio and HPE GreenLake cloud services. It offers consumption-based finance model like public cloud, but equipment is located on premises or wherever you choose.
So, when customers ask “what regions are you in” the answer is “in your region!” You can install your HPC infrastructure in your data center or a colocation of your choice. Wherever you are, you can build a private cloud service based on HPE GreenLake for HPC, or use a cloud service provided by a GreenLake for HPC deployment.
With HPE GreenLake, “HPC as a service” does not mean an HPC cluster in a public cloud provider’s setting. Nor does it mean HPE is building a public cloud platform. With HPE GreenLake for HPC, “HPCaaS” has taken on a new meaning. Infrastructure is not in the public cloud, but it is consumed in a similar fashion to public cloud.
So, why is it useful to consume on-premises (or colocated) HPC infrastructure as a service, as offered by HPE GreenLake for HPC?
Challenges solved by HPE GreenLake for HPC
Digital transformation became more urgent in the last two years, and public cloud adoption accelerated as businesses looked for the fastest way to meet new needs. Despite this, the vast majority of enterprise data still resides in data centers, colocations, and edge locations.
This is especially true for HPC workloads. Many organizations choose to keep their business-critical HPC workloads on-premises for a variety of reasons, including to meet data residency requirements or to ensure low-latency interactions between compute and storage.
At the same time, some want to consume HPC resources as a service, in order to simplify infrastructure and gain agility to scale quickly when their needs change. Hybrid cloud solutions, such as HPE GreenLake for HPC, have therefore emerged as the most practical way to run HPC workloads.
HPC infrastructure terminology
To help you understand these infrastructure options more clearly, we’ve defined some of the key terminology below with their specific objectives to meet your needs.
HPC as a service
HPC infrastructure is consumed as a managed service. The service is paid for on a subscription basis if reserved, or a measured usage basis if consumed on demand.
HPCaaS solutions can be delivered by a regional Service Provider or a Public Cloud Provider, which might provide managed infrastructure to process and store HPC data remotely. HPCaaS solutions delivered by HPE provide infrastructure at your data center, colocation or edge, with a cloud experience for finance, management, and deployment.
Hybrid HPC
When an HPC user wants to incorporate off-premise HPC environment into their HPC infrastructure, this becomes Hybrid HPC. There are different types of hybrid HPC where on-premise infrastructure and off-premise infrastructure used for internal or external workload mix: off-premise could be another site or facility or a campus; providing compute cycles to off-premises collaboration partners; or off-premise could be a collocated GreenLake for HPC infrastructure that is shared by multiple customers (we call it Shared Buffer) or off-premises could be a public cloud datacenter.
Selection of computing target is based on following attributes: scale, agility, cost, data locality, regulatory, capability matching, scientific software licensing constraints, real-time vs batch and external collaboration needs. Applications that will remain on premises are typically latency sensitive or those that are requiring local data processing or residency. An HPC user may want to choose an off-premise computing target if their compute demand occasionally exceeds both their reserved and variable compute capacity with their GreenLake for HPC solution. Users would like to have the same experience across on-premise, off-premise and public cloud: same reliable and secure HPC infrastructure, same services, tools (automation, deployment, security controls) and APIs.
Our partners offer a broad spectrum of software solutions and services that in combination with HPE GreenLake for HPC, provide a powerful end-to-end experience with the Hybrid model. As an example, we used Core Scientific’s Plexus to easily benchmark multiple public and private cloud service providers and compare price/performance between them; we also used Ansys Minerva to submit Ansys workflows on a different set of computing targets for finding best placement for a faster outcome.
Learn more about HPE solutions
HPE’s hybrid cloud HPC portfolio allows you to choose the best solution for every kind of need, whether your projects are in manufacturing, life sciences, healthcare, energy, or another science-focused industry.
Your HPC infrastructure is much more agile, because:
- You can right-size your solution and plan for future growth
- Buffer capacity is ready on site or off-premises
- You can scale up or down quickly to serve changing needs and new projects
- You can acquire any HPE HPC infrastructure with GreenLake – you choose whether to deploy a standard or custom solution.
The daily operation of HPC systems is also simpler. Provisioning HPC resources is fast and easy with the GreenLake Central portal. You can access all the professional, workload-specific services needed to achieve HPC goals for performance and resilience, as well as support to optimize workloads like containerization. And availability is protected, with downtime-free maintenance that leverages your capacity buffer
To learn more about your options for meeting HPC infrastructure challenges, read the resources below or speak to your HPE representative.
Max Alt
Hewlett Packard Enterprise
twitter.com/hpe_hpc
linkedin.com/showcase/hpe-ai/
hpe.com/info/hpc
Max_Alt
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. He 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 areas of expertise lie in computer and server architectures, cloud technologies, operating systems, compilers, and software engineering. Prior to joining HPE, Max was SVP AI & HPC Technology at Core Scientific that acquired Atrio in 2020 where Max was the CEO & Founder.
