top of page
AUTOMATION:
World leader in telecom and wireless technologies
01
Context:
The customer wanted to transform its bare metal servers into a Multi-access Edge Computing (MEC) environment in order to optimise the GPU/FPU consumption of its mobile gaming applications.
02
Problem solving approach:
Develop a containerised approach, implement a global and fully automated architecture to transform the bare metal GPU/FPU servers into on-premises Kubernetes clusters.
03
Result:
Successful demonstration of GPU slicing and optimisation (Nvidia Grid) using the new architecture and approach to application deployment (between 30 and 40 GPU/FPU nodes per bare metal server, transformed into a K8S cluster).
04
Technical stack used :
Kubernetes, Kubespray, Ansible, Jenkins, Redhat Virtualisation/Ovirt
bottom of page