Elastic scaling of PCI devices in a private cloud platform.
LONDON, United Kingdom. – Aug 13th, 2020 – vScaler, an opensource Private Cloud offering, built and designed by HPC experts for HPC workloads, today announced the integration of SLURM with GigaIO’s FabreX offering within its Cloud platform, enabling elastic scaling of PCI devices and true HPC disaggregation.
As the industry’s first in-memory network, FabreX supports vScaler’s private cloud appliances for workloads such as deep learning, biotechnology and big data analytics.
vScaler’s Disaggregated (also known as ‘Composable’) HPC solution is enabled by FabreX’s Cloud-class infrastructure platform, a PCIe network that allows users to create rack-scale servers to drastically improve the utilisation rate of expensive resources like GPUs and FPGAs as they can be reconfigured on the fly as workflows change and evolve.
vScaler Chief Technology Officer, David Power comments “We’ve integrated GigaIO’s FabreX hardware into our private cloud product to allow our customer dynamically reconfigure hardware in line with evolving user demands and workloads. As people start to employ more AI tools and techniques, the underlying requirements for hardware to accelerate those tools also evolves. FabreX allows us provision resources to workloads so that we can run much more diverse workloads on top of a core hardware platform”.
“FabreX is based on an open architecture offering industry standard Redfish APIs that make it easy to setup, and because we partner with top tier providers like vScaler for seamless integration, we’re delivering true cloud-class orchestration and composition capabilities,” said Alan Benjamin, CEO of GigaIO. “vScaler’s SLURM integration with FabreX enables end users to dynamically compose their own infrastructure by reaching inside the rack to create servers and resources in seconds, complete with leading security, access control and provisioning features, to match the needs of today’s workflows used in cloud deployments while optimising TCO.”
The additional integration of the SLURM workload manager, an open-source job scheduler for Linux and Unix-like kernels, means that vScaler Cloud users can request traditional resources like memory and compute cores to be available for jobs. Now, coupled with FabreX, users can now also specify PCI devices such as NVMe or GPUs to be attached to cluster nodes when running and executing specific workloads.