As more and more HPC users begin to look to the cloud to help solve their workload challenges, we have decided to run a series of articles introducing readers to this emerging landscape. While we usually post articles more technical in nature that is not our intent here. In fact, our aim is to demystify HPC migration to the cloud for people who are thinking about or considering cloud for the very first time. We hope these articles and resources will prove helpful as you navigate the promising new world of cloud-based HPC.
What is Cloud-based HPC?
At a basic level, cloud computing is the delivery of services over the Internet, such as servers, storage, databases, networking, software, and of course computing, by providers who charge for access based on customers’ usage. The majority of these services fall into three general categories: software as a service (SaaS), infrastructure as a service (IaaS), and platform as a service (PaaS).
Most people are familiar with SaaS, where providers host, manage and deliver software applications, such as Office 365, Google Apps, and Salesforce.com, over the Internet to their subscribers.
Similarly, PaaS is an on-demand environment for developing, testing, delivering, and managing software applications, making it easy for developers to create applications without having to set up and manage the underlying platforms they require.
With IaaS, a customer ‘rents’ infrastructure, such as virtual machines, servers, storage, networks, operating systems and other infrastructure software from a cloud provider on a pay-as-you-go basis. IaaS is an alternative to provisioning, installing, and maintaining a large part of your IT infrastructure.
Many organizations realize tremendous advantages using public cloud over in-house infrastructure, as they shift what was a large capital expense to an operational expense, while gaining immediate availability and rapid scalability and access to specialized resource. While cloud computing has been widely adopted by companies in various forms, until relatively recently it wasn’t a viable option for many large enterprise HPC sites. But times have changed!
Current market adoption of cloud HPC
As HPC workloads become more compute-intensive, there is a corresponding demand from researchers and developers for ever more computing power. Market estimates point to a recent surge in cloud spending by HPC customers. According to Intersect360, high-performance cloud spending by HPC customers grew by 44 percent from 2016 to 2017, reaching about $1.1 billion US. Indeed, our own recent customer survey found that adoption of cloud among our own HPC customers is clearly on the rise.
Why are enterprises now moving HPC to cloud?
A confluence of factors is now driving cloud computing adoption by HPC users. Today, newer hardware in public clouds offers tremendous improvements over aging on-premises infrastructure, especially for customers running (or wanting to run) machine learning applications, where the latest GPUs provide the superior capabilities needed for those types of workloads. HPC users are also attracted to the cloud’s new, advanced offerings: applications, developer tools, virtual private networks, big data analytics, management software, and so much more.
The case for cloud-based HPC is driven by various business motivations. Here are a few:
A growing demand for HPC, and the time and expense required to deploy and manage on-premises HPC infrastructure, is leading many users to consider cloud, either to augment their existing infrastructures (i.e. hybrid cloud), or to entirely replace them (i.e. dedicated cloud). If you are beginning to explore the potential of cloud for your HPC environment, we encourage you to follow this blog and if you have questions or would like to speak with one of our product experts, please email us at firstname.lastname@example.org