Missed a session at the Data Summit? View on demand here.
Rescale, a high-performance cloud computing provider, has unveiled new features aimed at improving container support across cloud services and specialized architectures. The new offering should simplify the development of advanced scientific and technical applications for digital twins and advanced scientific research.
Containers are one of the fastest growing sectors in the IT industry. Containers help lower the barriers when moving applications from on-premises to cloud infrastructure or between clouds. Gartner predicts that more than 75% of global organizations will have containerized applications in production this year, up from less than 30% in 2020.
However, businesses tend to face more challenges in moving high-performance computing (HPC) workloads to containerized infrastructure. HPC use cases have additional security and parallel processing requirements. Rescale’s special sauce includes bundles of HPC-specific apps and the dependencies needed to run the best architecture possible.
Simplifying the engineering experience
Rescale provides a user-intuitive interface for engineers and scientists to securely deploy and scale containerized workloads on any architecture from a multi-cloud infrastructure catalog. The company is now expanding support for HPC use cases around artificial intelligence (AI) and machine learning (ML) workloads. Containers allow engineers and scientists to pull published containers from their registry of choice and continue their research without thinking too much about the infrastructure.
Containers are useful for deploying specialized tools, such as an Nvidia solver that specializes in their architecture or in-house developed technical applications. With Rescale, users can pull published containers from their registry of choice and continue their research without thinking about the infrastructure. Rescale automates the entire workflow to run these applications and provides users with easy collaboration to share tasks or data, while providing IT/HPC organizations with security, financial and architectural controls.
This simplifies the architectural decisions needed to build the right infrastructure around any application the user wants to use. The company also has unique information about the full set of cloud provider architectures. This includes how each architecture performs against different applications at different scales, the maturity and available capacity of these architectures in different geographies and regions of cloud providers. This can range from a workload for AI/ML, computational fluid dynamics and computational chemistry. This has helped enterprises such as nuclear fusion reactors (Commonwealth Fusion Systems) and commercial spacecraft (Firefly Aerospace).
Refactoring for cloud services
Adam McKenzie, CTO and founder of Rescale, told VentureBeat Rescale has focused on two key drivers for restructuring existing HPC workloads and tools to run on microservices in the cloud infrastructure.
First, a lift-and-shift approach won’t work for many HPC applications because of the gravity challenges, network bandwidth, and complex infrastructure configuration. Rescale rearranges these components into a software-as-a-service offering. Rescale maintains all software that an engineer runs from the catalog. Companies can also bring their software and publish it in the catalog and Rescale optimizes the architecture for different platforms.
Second, Rescale tracks cloud offerings, pricing, and performance changes from cloud providers. This helps ensure that each application uses the best architecture, runs in the best regions, and pays for the right level of service.
McKenzie said, “This cloud-built approach is reminiscent of how Snowflake reimagined what a user’s data warehouse experience in the cloud should be. What Snowflake did with the data warehouse, Rescale is doing with the accelerated compute stack.”
VentureBeat’s mission is to be a digital city square for tech decision makers to learn about transformative business technology and transactions. Learn more