Choosing the Right Cloud Infrastructure for the Industrial Metaverse

Imagine stepping into the Industrial Metaverse, a world where time and space take on new dimensions. Here, businesses and organizations have the keys to transformation, reshaping the way decisions are made.

In this virtual world, you aren’t just a spectator, you’re a participant, directly manipulating your environment with a power that was once thought to be fantastical.

But such power relies on a strong foundation — the cloud infrastructure. Choosing the wrong infrastructure leads to performance issues, creating delays and resource bottlenecks that can disrupt collaborative simulations and increase costs.

In this article, we're exploring the criteria businesses should use for picking the proper environment for Industrial Metaverse. This includes the need for substantial resources, robust rendering, collaborative content creation, and real-time processing. We’ll focus on teams engaged in heavy-load, real-time ray tracing visualization, as well as automotive design and engineering teams.

Choosing the most suitable Metaverse environment involves a combination of both objective and subjective criteria.

1. GPU Availability

To qualify as an Industrial Metaverse, it’s crucial to have a variety of virtual machines equipped with different Nvidia Graphics Processing Units (GPUs). This diversity allows users to pick the most suitable option, whether they are looking for cost-effectiveness or better performance in tasks like rendering and computations.

While regular Central Processing Units (CPUs) work well for general tasks, GPUs shine in specific jobs such as scene rendering, artificial intelligence (AI), and simulations on a large scale. Nvidia has made this even more specific by creating various families of server GPUs for different tasks.

Our evaluation of different platforms revealed a disappointing lack of variety in GPU offerings, underscoring the importance of having a range of GPU-based virtual machines to cater to varying needs. Whether you're testing 3D models, running simulations, creating realistic images, or building a virtual development environment, having diverse GPU options is vital.

2. Price per FPS

While the whisper of cost-effectiveness might tempt you to make your selection based solely on price-per-environment, the true economical choice whispers in the language of Frames-Per-Second (FPS). As we found out, this productivity metric may not be loud, but it speaks volumes about the true cost-effectiveness.

Taking a particular cloud provider as an example, we compared rendering and FPS to illustrate the importance of this criterion. To measure FPS, we utilized visual demos from SoftServe and Nvidia. Some virtual machines (VMs) faced challenges in running the Nvidia demo, making it difficult to reach a definitive conclusion. However, for those where it was possible, we calculated the average FPS per USD after comprehensive measurements on both demos. Additionally, Hyperstack by NexGen Cloud was used as an environment, offering a commendable variety of Nvidia GPU VMs.

Exploring the possibility of choosing a virtual machine with multiple GPUs adds an intriguing layer to the pricing analysis. The overarching message is clear: considering the price per FPS can guide you toward your solution goal with reduced expenses.

3. Environment Onboarding Ease

While quantitative measures are valuable, the user experience plays a crucial role, impacting how effectively users can utilize the environment.

This setup process starts from the homepage and ends with gaining final access to the environment. Given the research and development nature of the relatively new Industrial Metaverse field, there's a continuous cycle of prototyping, altering, and re-prototyping. In such scenarios, the speed with which one can establish a new environment becomes vital.

As different cloud providers offer a diverse range of GPU environments and other features, we found it impossible to use time as a measurement criterion. Consequently, we decided to keep this criterion subjective.

Although no cloud provider currently offers a one-click setup, some provide a more user-friendly option, enabling the setting up of an environment with just a few clicks and featuring a less complicated user interface and user experience. Therefore, users should form their own opinion on how easy it is for you to set up and access a new GPU-based virtual machine.

A few pointers to look for in a suitable environment:

When choosing a cloud infrastructure for your Industrial Metaverse project, consider your specific requirements. Are you prioritizing cost-effectiveness, high performance, or a specific type of GPU? Is the speed of setting up an environment essential for your research and development activities?

These considerations often lead to practical criteria derived from implementing numerous solutions. By assessing your needs and comparing offerings from various providers, including hyperscalers and specialized GPU cloud platforms, you can select the cloud infrastructure that best supports your Industrial Metaverse journey.

If you need help customizing Industrial Metaverse solutions, SoftServe is here to support you with cloud deployment using Nvidia Omniverse. This deployment includes a self-healing infrastructure and scalable on-demand features, making it easy for globally distributed teams. We assist in setting up Omniverse Enterprise, creating custom extensions to enhance capabilities, and seamlessly integrating with leading industry software applications.

Let's talk about creating the Industrial Metaverse solution that precisely fits your needs.

Disclaimer: Conducted in fall 2023, this SoftServe’s research captures a snapshot of the rapidly changing GPU landscape. As the Industrial Metaverse evolves, the criteria, mentioned above, most likely will change as well.