AWS Startups Blog

Turn it on, scale it up: OnScale CEO Ian Campbell on on-demand CAE with infinite, scalable cloud HPC

Guest post by Ian Campbell, CEO of OnScale

The core team at OnScale. Photo: Vicki Thompson

I’m a hardware junky. I’ve got an oscilloscope, a function generator, several multi-meters, and boxes full of circuit boards, chips, wires, LEDs, Arduinos, and Raspberry Pis… and that’s just in my home office! In the lab at my first startup company, NextInput, we had every tool imaginable to design, manufacture, test, and deploy sensors and sensor systems for applications like smartphones and Internet-of-Things (IoT) devices. We loved our engineering tools.

One set of engineering tools I always hated, though, was legacy computer-aided engineering (CAE) software tools and the old, worn-out high-performance computing (HPC) hardware we ran our CAE tools on. Legacy CAE tools are expensive, slow, inaccurate, and never really gave us the engineering insights we were looking for when optimizing sensor and system designs—huge, CPU and RAM intensive problems. I thought there had to be a better way.

Turn it On, Scale it Up: On-Demand CAE with Infinite, Scalable Cloud HPC

At OnScale, our team has a very clear mission: break compute and cost constraints for engineers using CAE tools to design and optimize next-generation products. We did that by moving the CPU and RAM intensive components of CAE tools—the multi-physics solvers—to the AWS Cloud.

Rather than haphazardly creating AWS Cloud HPC instances of fixed sizes (for example, the fixed number of CPUs and RAM), we used a Docker container approach to deploy small software containers that include our powerful OnScale multi-physics solvers and pre-/post-processing tools. These completely independent and secure containers run on the newest class of AWS HPC Instances and use just enough CPU power and RAM to complete a given engineering job. This means that we’re not wasting a single core-hour of CPU power and not a single MB of RAM. To the end customer—engineers—the user experience is seamless. It’s as if they are running analysis on their local HPCs, but only much, much faster.

Engineers don’t care where or on what type of HPC their analysis is run, they care about getting results back. Fast. With OnScale, engineers can run massive studies and massive numbers of studies in parallel and get the results they need back 10 or even 100 times faster than using Legacy CAE/HPC. Since our OnScale solvers are running on the AWS Cloud, we can offer them to engineers through cost-effective monthly or annual subscriptions. This is how we’re breaking both performance and cost constraints for engineers.

Always Available, Secure Architecture

When a local HPC goes down at an engineering firm, all productivity grinds to a halt until it’s back online. In some cases, this could take days or even weeks—it’s a productivity killer. And in fast-moving highly competitive industries, HPC downtime burns cash, not just productivity.

With OnScale and our AWS Cloud HPC platform, we can offer engineers access to world-class CAE multi-physics solvers with an “always-on” availability. The system is always up, waiting to crank through the toughest engineering problems. Engineers will no longer have to wait for IT to set up local HPCs, distribute and manage software licenses, or provision data storage.

All data is end-to-end encrypted using the AWS Cognito Identity Provider and Key Management System. We like to tell customers “It’s your Cloud”—because it truly is! All user management, encryption key management, and data transfer happen solely between our customers and AWS—OnScale isn’t in the middle of those transactions and even we can’t look at our customers’ encrypted data at any point in the process.

The Future of Engineering is Here

OnScale, powered by the AWS Cloud, will forever change engineering and open up a new wave of innovation in fast-moving, high-tech sectors. We’re very excited to be working with AWS to bring the Future of Engineering to every engineer.