In today’s environments, IT organizations need to support growing numbers of artificial intelligence applications, along with many other new and emerging high performance computing workloads. This mission creates challenges that aren’t necessarily solved with conventional IT technologies and approaches. In a typical scenario, IT administrators set up separate systems to accommodate AI workloads, and then manage many aspects of those systems with time-intensive manual processes.
Today, there is a better way forward — thanks to the integration of technologies from VMware, NVIDIA and Dell Technologies. With NVIDIA and VMware collaborating on software, IT organizations can run traditional and AI workloads in the same environment and on the same systems. This makes life easier for the people tasked with delivering the information technology to support emerging workloads. AI applications can now be managed with the same VMware flexibility as with other applications.
In addition, with the tight integration of VMware and NVIDIA offerings, organizations can now virtualize multiple technologies inside their systems. For example, they can virtualize and share the GPUs inside servers to enable multiple data scientists to simultaneously accelerate their deep learning workloads. This increases utilization rates and saves money that would have otherwise been spent on the procurement and management of additional hardware.
This integration of technologies also helps IT organizations save time and administrative steps. And perhaps best of all, they can manage everything centrally with VMware vCenter and easily allocate resources as needed.
These are the kinds of benefits that the University of Pisa is realizing today with its use of Dell Technologies with VMware and the NVIDIA AI Enterprise software suite.
Solution highlights
- Dell EMC VxRail provides a simple, cost effective hyperconverged infrastructure that solves a wide range of IT challenges and supports almost any use case, including tier-one applications and mixed workloads. VxRail enables faster, better and simpler delivery of VMware-virtualized applications. For a seamless, curated and optimized HCI experience, VxRail is engineered jointly by Dell Technologies and VMware. VxRail is also an NVIDIA-Certified System, which means it has been validated to provide excellent performance, security, and scalability for AI and data science workloads.
- Dell EMC PowerScale storage is designed to serve as the foundation for data, building an integrated and optimized IT Infrastructure for AI initiatives, from proof of concept (POC) to production. These all-flash scale-out network-attached storage solutions deliver the analytics performance and extreme concurrency at scale to consistently feed data-hungry deep learning algorithms. And combined with PowerScale OneFS governance and enterprise features for data management, data security, data compliance and data protection, PowerScale storage helps IT organizations conform to regulatory and enterprise security policy requirements.
- NVIDIA AI Enterprise is a software suite of enterprise-grade AI tools and frameworks that is optimized, certified and supported by NVIDIA with the latest VMware vSphere. With this software, IT professionals at the hundreds of thousands of enterprises that use vSphere for compute virtualization can now support AI with the same tools they use to manage large-scale data centers and hybrid cloud environments. NVIDIA AI Enterprise provides scale-out, multi-node AI application performance on vSphere that is indistinguishable from bare-metal servers.
- VMware vSphere is the industry’s leading server virtualization software for applications using any combination of virtual machines, containers and Kubernetes. Rearchitected with native Kubernetes, you can now modernize the 70+ million workloads running on vSphere. And now, you can run modern, containerized applications alongside existing enterprise applications on existing infrastructure with vSphere with Tanzu.
A Center of Excellence
The University of Pisa is both a Dell Technologies and a VMware AI Center of Excellence. As part of this designation, the University’s IT team regularly evaluates and tests new technologies. That’s the case today with the combination of Dell Technologies with VMware and NVIDIA AI Enterprise.
“We are running AI workloads on top of VMware, and we are using Dell EMC PowerStore for the storage in this virtualized environment,” explains Maurizio Davini, chief technology officer for the University of Pisa. “And we have a Dell EMC PowerScale all-flash environment for AI and HPC as a sort of traditional scale-out in our fast systems.”
Davini notes that his organization has both bare-metal and virtualized systems with NVIDIA GPUs and DPUs.
“We have traditional bare-metal GPUs, which are typically used for research like language processing, image processing, deep learning, deep neural network research and so on,” he says. “So we are still increasing our bare-metal capabilities on GPUs. And we now also have clusters of GPUs inside our VMware production environment which match the performance level of our bare metal systems.”
Flexibility is the key here.
“VMware gives us the possibility to be flexible and to use the infrastructure for a lot of things — enterprise workloads, VDI, remote workstations, support for smart working, scientific computing, HPC — all in the same infrastructure in a very flexible way,” he says. “And this is the problem that VMware and Dell have helped us to solve.”
For the full story, see the Dell Technologies case study “Simplifying AI systems.”