Edge Cloud Platform Development for a German Research Organization
Architecture and development of distributed edge cloud platforms based on OpenStack and Kubernetes.
Company
German · 500 employees
The client is a German research organization with less than 500 employees. Its infrastructure includes multiple distributed edge cloud platforms based on OpenStack and Kubernetes. The project also involved networking, Kubernetes-based environments, and NVIDIA vGPU support as part of the organization’s edge cloud infrastructure.
Challenge
The project involved the architecture and development of several edge cloud platforms based on OpenStack and Kubernetes. The scope covered both cloud infrastructure and networking, as well as the deployment and evolution of individual edge nodes.
KEY CHALLENGE
Building and evolving multiple edge cloud platforms with reliable networking, automated switch failover, and support for NVIDIA vGPU.
Project specifics
The project combined the development of several OpenStack- and Kubernetes-based edge cloud platforms with extensive work on the underlying network infrastructure. This included designing the leaf-spine architecture, implementing automatic switch failover, configuring firewalls, updating Kubernetes releases, and deploying new edge nodes from scratch.
Solution
Cloudification worked on the architecture and implementation of the platforms, including the design of a leaf-spine network architecture and automatic failover in case of a switch failure. The team also handled Kubernetes release upgrades, deployment of new edge nodes from scratch, firewall configuration, and migration of the base operating system from old unsupported CentOS version to Ubuntu LTS. End-user training was also provided as part of the project.
Result
10+ edge cloud platforms based on OpenStack and Kubernetes were deployed with the required network architecture based on Cumulus Linux with BGP and BFD. The infrastructure was extended with new edge deployments, while the migration from CentOS to Ubuntu LTS allowed the integration of Nvidia vGPUs. End users also received training on working with the implemented infrastructure.
Private Cloud Platform for IoT Infrastructure
OpenStack and Kubernetes environment with Gardener and Keycloak integrations, NVIDIA L40 vGPU capabilities, and administrator training for the customer’s internal team.