VMware has introduced a new integrated feature in VMware vSphere 7 that will enable enterprises to deliver elastic infrastructure on-demand for artificial intelligence (AI) and machine learning (ML) applications.
This new feature—VMware vSphere Bitfusion—is developed out of VMware’s 2019 acquisition of Bitfusion, a pioneer in the virtualization of hardware accelerator resources including graphics processing unit (GPU) technology.
The combination of Bitfusion and VMware vSphere will help organizations achieve cost savings, enable resource sharing out of the box, and deliver the right hardware accelerator resource, like a GPU, to the right workload at the right time.
“We aim to deliver the same value to GPUs that we delivered for CPUs,” said Krish Prasad, senior vice president and general manager, Cloud Platform Business Unit, VMware.
“By breaking down existing silos of GPU resources, organizations will be able to achieve better utilization and efficient use of them through sharing—resulting in immediate cost savings. More importantly, organizations will be able to jumpstart new or stalled AI/ML initiatives to drive their business forward by sharing those GPU resources with their teams on-demand with VMware vSphere 7.”
Efficient GPU Pooling and Sharing
AI and ML-based applications—deep learning training in particular—rely on hardware accelerators to tackle large and complex computation. With the newly integrated Bitfusion capabilities, VMware vSphere 7 will enable enterprises to pool their powerful GPU resources on their servers and share them within their data centers.
That will enable organizations to efficiently and rapidly share GPUs across the network with teams of AI researchers, data scientists and ML developers relying on and/or building AI/ML applications.
Released in April 2020, VMware vSphere 7 was rearchitected into an open platform using Kubernetes to provide a cloud-like experience for developers and operators. The Bitfusion feature of VMware vSphere 7 will leverage GPUs for applications running in virtual machines or containers. Bitfusion can operate in a Kubernetes environment such as VMware Tanzu Kubernetes Grid, and is expected to run side-by-side as customers deploy AI/ML applications as part of an overall modern applications strategy.
The Bitfusion feature of VMware vSphere will be available through a single download with no disruption to current infrastructure and will seamlessly integrate with existing workflows and lifecycles.