For more than 30 years, the Ohio Supercomputer Center has been on a mission to make high performance computing (HPC) resources and expertise available to university and industrial researchers in Ohio and beyond. Whether researchers need to harness the power of a parallel-processor cluster to better understand deep space, a vector processor machine to do weather modeling, or a midsize shared memory processor system to model the human heart, OSC has the systems to meet their needs.
In addition to providing these leading-edge computational solutions, OSC offers Open OnDemand, a portal that simplifies access to HPC resources. This web-based portal, built on open-source software developed at OSC, serves as a “one-stop shop” for accessing HPC from anywhere. With Open OnDemand, teams can upload and download files; create, edit, submit and monitor jobs; run GUI applications; and more via a web browser, without client software to install and configure.
Open OnDemand removes many of the complexities that come with accessing HPC systems by traditional methods, according to Eric Franz, a lead engineer for the Open OnDemand team at OSC.
“The biggest benefit of Open OnDemand is reducing the time to science,” Franz says. “Without OnDemand, if you’re a researcher or a student coming to OSC for the first time, you might have to install on your laptop a terminal emulator and a file-transfer application, and you’re going to have to do a lot of learning on how to craft a batch script in order to submit jobs on our clusters. But with on Open OnDemand, all you need is a URL, a username and a password, and you have access to our clusters. And so Open OnDemand makes it very easy for researchers and students to get started with HPC.”
Sharing the goodness
Open OnDemand is part of a trend toward operating HPC resources as cloud services with clean web-based interfaces and well-defined service-level agreements. Increasingly, systems that at one time might have been locked up and made available to only select power users are now within the reach of many users via cloud interfaces.
Today, more than 120 research institutions around the world are offering easy, web-based access to HPC systems via Open OnDemand. Among them is the University of Florida. With Open OnDemand, researchers there can access the UF HiPerGator supercomputer with a simple web browser, according to Dr. Erik Deumens, director of research computing for the University of Florida.
“With Open OnDemand, all the intense computation is being done on the server,” Deumens says in a Dell Technologies case study. “The only thing that is sent to the browser are the actual pixels that the user sees. You can do it even when you are connected on a relatively poor network connection. You don’t have to be in a super fancy room with a high-performance network connection.”
While saving time and simplifying life for seasoned users of supercomputers, Open OnDemand simultaneously opens the door to HPC to new users.
“Open OnDemand enables users who might otherwise be turned off to HPC because it’s too difficult, Franz says. “It makes it easier for new disciplines to take advantage of HPC, which is very important because we now have a lot of big data problems that we didn’t have in the past. Many non-traditional fields now have HPC problems.”
Franz notes that there’s nothing that Open OnDemand can do that individuals with a terminal emulator couldn’t do themselves. It’s just that with a terminal emulator, it might take 20 or 30 steps, along with some systems expertise, to get a job up and running in a supercomputer. Today, thanks to Open OnDemand, users can accomplish the same steps with the click of a button.
“With Open OnDemand, researchers in many fields can get their results faster,” Franz says. “And it’s a heck of a lot easier than it was before.”
For the full story, including a look at the use of Open OnDemand to enable COVID-19 research, see the Dell Technologies case study “Making HPC ‘Open OnDemand’” and watch the video.