Cloud computing is more prevalent today than it was some years ago, but bigger companies still face compatibility and scalability issues.
A cloud platform being not compatible with various technologies and software is a problem that many face, but it especially affects the field of engineering. Many engineers have to run countless complex simulations that require high computational power. And while the cloud may be faster than an on-premises environment, if it’s not perfectly tailored to a particular use case, the results might be even worse.
To discuss the importance of a highly adaptable cloud computing environment and the role of cloud experts in organizations, Cybernews reached out to Wolfgang Gentzsch, the President of UberCloud, which is a cloud simulation platform for engineers.
How did the idea of UberCloud originate? What has the journey been like since your launch in 2012?
The journey has been phenomenal, so far. Before we started UberCloud in 2012, I have been an engineer, scientist, professor, entrepreneur, and industry consultant, always with a focus on high-performance computing (HPC), for super-, vector-, parallel-, and grid-computing infrastructures applied to complex engineering simulations. I worked with exceptional engineers in mechanical, electrical, and aerospace engineering. I know full well that how innovative and productive these amazing brains can be is limited by how much computing power is within their reach. I feel a responsibility to make it easier to get compute power in the hands of amazing engineers because this is how humanity moves forward.
When I first met my UberCloud partner Burak Yenier in April 2012, he was the VP of Operations at CashEdge, now part of Fiserv, responsible for the data center and payments infrastructure of the financial services technology company. Already then, Burak had profound expertise in cloud computing. It quickly became clear to us that our complementary expertise was an ideal mix of cloud computing, engineering simulations, and High Performance Computing (HPC), and we recognized that cloud computing could potentially become the next big thing for complex engineering simulations. In those early days, these simulations were mainly running on very limited on-premise computing systems. What if we built a fully automated, on-demand, self-service environment for engineers? And empower them to perform their simulations on-demand, on much faster, many more, latest-technology HPC resources? And help them accelerate their next-gen product developments, by a large factor? And that’s what Burak and I put into practice then.
Fast forward to 2022 and we are proud to see millions of CPU core hours of compute capacity flowing through the UberCloud platform to brilliant engineers and to be part of the innovation they produce that makes our lives better.
Can you tell us a little bit about your HPC Cloud platform? What are its key features?
Our cloud platform today consists of two major components: first, UberCloud’s fully portable Engineering Simulation Platform provides an automated self-service environment for engineers. With a few clicks, they can set up their own custom computing environment on top of any hybrid or public cloud infrastructure, perfectly tuned to their specific application workflows. The second component is our standard UberCloud HPC containers hosting the engineers’ complex simulation workloads. These HPC containers are built on standard micro-services container run-time technology and enhanced for what we call “macro-services” by integrating out-of-the-box HPC features like a Message Passing Interface (MPI) and Infiniband Interconnect (for multi-container multi-node scalability), GPUs (for accelerating computing), batch and interactive processing, storage connectors, a software license server, container security (in addition to cloud providers’ own security stack), batch scheduling, and even a fully equipped virtual desktop.
The UberCloud platform is a no-compromise solution: we provide mechanical, electrical, and aerospace engineers with unlimited computing power to run their favorite engineering applications such as ANSYS, Dassault, Siemens, and many many others. In the meantime, the UberCloud platform runs natively on any computing platform the IT team wants, such as popular public Cloud offerings of Amazon, Microsoft, and Google. In summary: full control for IT and a lot of simulation features and scalability for engineers.
What would you consider the main challenges engineers run into nowadays?
Long job queues! Engineers are used to running their most complex, often multi-physics, many-hour simulation workloads on-premises, sitting in long waiting queues, and competing with jobs from their colleagues, for limited compute resources that age quickly over their 5-year lifetime and have long procurement cycles. Often enough, engineers have to perform hundreds of different design simulations, with many parameters, different geometries, and physics conditions, for the next-gen product that should have greater features and higher quality than its predecessor. All these on-premise limitations that I just mentioned are dramatically slowing down innovation, competitiveness, engineering productivity, and time to market.
How did the recent global events affect your field of work?
We saw a major shift recently towards working from home. Engineers who depend on their physical workplaces are severely threatened by these global challenges and companies are looking for alternatives. Our customers are developing their next-generation products on UberCloud’s automated and self-service engineering simulation platform and are thus location-independent even during such global events mentioned previously.
What are the best practices companies should follow when developing, and, when launching applications?
Here we have to differentiate between small to medium-size companies (SMEs) and large enterprises. SMEs don’t usually develop their own applications. They use off-the-shelf commercial simulation software from the likes of Ansys, Dassault, Siemens, and many other engineering software providers. UberCloud provides all this software packaged in containers on top of its simulation platform, ready to work, no learning needed, at their fingertips. For large enterprises, on the other hand, containerized commercial software provides big benefits as well, but also and especially for their own proprietary in-house software developments. In a hybrid on-prem/cloud environment, this software can be developed on workstations, or on-premises, but tested, certified, and launched in the cloud, on much larger and more powerful HPC systems.
As more companies move their workload to the cloud, are there any details that might be overlooked when making the switch?
The major misunderstanding that we recognize from companies that first contact us is that they expect to reduce costs. And then they compare their on-prem cost with pricing in the cloud, often missing to include all the on-prem direct and indirect costs into their calculation, while for the cloud their math is usually based on expensive on-demand compute instances, ignoring the different discount opportunities that they can get when they commit to a certain time and resource consumption, and additional discounts for reserved and spot instances. But there are other benefits of using the cloud that are by far more important than just the cost.
Today, we can keep the cost of cloud in the same range as their on-prem computing cost, but in addition, an automated cloud platform comes with higher agility, much easier innovation enabling, increased engineering productivity, product quality, competitiveness, and more. One of our recent customers, Rimac Technology, the builder of the fastest electric hypercars in the world, has recognized this early and found the perfect balance between strategic benefits and cost savings.
The second major misunderstanding is that do-it-yourself (DIY) can save a lot of money compared to working with external HPC cloud experts that have already successfully migrated other companies’ workloads. DIY projects, however, often fail for several reasons. One, the cloud is a C-level responsibility and requires C-level commitment. Only then can early internal opposition be circumvented. Without such a commitment, a cloud project can slow down and die quickly. Second, a company’s traditional compliance rules have to be adjusted to accommodate outsourcing requirements like moving workflows and data to a public cloud. Without compliance adjustments to the cloud, we have seen lengthy internal disputes resulting in delays of many months to even several years. Third, cloud acceptance exploded over the last few years, and there are not enough cloud experts available in the market who mostly have been hired away already by the big cloud providers. Fourth, just moving applications ‘manually’ to one specific cloud infrastructure doesn’t mean that it works “out of the box” on other cloud infrastructures. Usually, the application has to be ported from scratch. And when a new software release is available, part of this porting has to be done again, often three times a year, for each new release. And fifth, to support DIY cloud activities, a team of internal cloud support experts is necessary. Such an expert team is expensive, more expensive than any professional (and automated) cloud platform software. And these experts can leave the company all of a sudden.
What types of attacks do you think are going to emerge as a result of the growing use of the cloud and other advancing cybersecurity measures?
Cloud providers like Microsoft are spending billions of dollars each year on cybersecurity. For an individual company, such a level of security is difficult to reach and maintain. And if someone wants to steal a company’s assets, it’s easier to focus directly on the company’s own IT infrastructure than on the cloud with its thousands of users from thousands of companies which is like looking for a needle in a haystack. The vulnerability of a company’s assets comes often from its own employees’ thoughtless use of login information, passwords, data, and other assets.
In your opinion, what cybersecurity best practices every organization should follow to combat emerging threats?
As previously mentioned, public cloud infrastructures, especially from leading cloud providers, are as secure as Fort Knox (smiling). And with UberCloud’s secure application containers on top, this security level will certainly be maintained. The weakest link of the security chain is the human in the middle. Thus, the handling of security information (including login and password for the company’s cloud environment) has to be baked into the company’s cloud policy. And the handling of confidential information and processes has to be clearly specified and employees have to be trained and reminded regularly.
What does the future hold for UberCloud?
To the best of my knowledge, the UberCloud Engineering Simulation Platform together with our unique HPC containers is the most flexible and portable cloud services technology in today’s market. Our customers large and small help us improve our products every day just by using our technology and providing valuable feedback. And because this cloud software is a horizontal technology that is cloud and application-agnostic, it not only supports software for computer-aided engineering (CAE), life sciences, finance, energy, and more, but also any compute and data-intensive application can dramatically benefit from this cloud technology, as we have demonstrated with several UberCloud Experiments. Such cloud technology may include Multi-Physics Simulations, Construction, AI/Machine Learning, Digital Twins, Data Analytics, Personalized Healthcare, Medical Devices, Natural Language Processing, Autonomous Driving, and IoT/Edge in the near future.