/nfv-k8s-cluster

This project hosts resources to create an NFV Kubernetes Cluster to deploy cloud-native VNFs

Primary LanguageShell

NFV Kubernetes Cluster

This project hosts resources to create an NFV Kubernetes Cluster to deploy cloud-native VNFs.

This is still work in process at present

Infrastructure

A basic set of tools allow to provision the underlying infrastructure in different cloud provider environments. The NFV cluster is deployed into the created infrastructure. The infrastructure provisioning is based on Terraform.

At present, the following providers are supported:

  • OpenStack
  • WiP Digital Ocean

Requirements

  • terraform binary (see Install manual)
  • API keys for the respective cloud provider

Find the supported infrastructure providers in infrastructure.

  • cd infrastructure/openstack
  • Create a file terraform.tfvars based on the given example
  • Run terraform init to initialize terraform for your project
  • Run terraform plan to see the execution plan
  • Run terraform apply to execute the plan an provision the infrastructure

Cluster Setup

A Kubernetes cluster is created on a given infrastructure. Besides the fundamental Kubernetes components, additional resources are created to run VNFs. The cluster boot-strapping utilizes kubespray.

Quick-start

Next step is to use kubespray+ansible to deploy k8s onto the infrastructure. (Example steps below are based on openstack example.)

  • Run ./run-kubespray.sh (This will clone kubespray repository, create ansible-inventory from terraform-state and run the playbook for k8s deployment)
  • Run ./describe-inventory.sh terraform.tfstate to see infrastructure details.
  • Copy /etc/kubernetes/admin.conf from a master node.

To connect pods to multiple interfaces using multus CNI, adapt the CNI configuration 05-multus.conf to your needs and deploy it on the respective hosts into /etc/cni/net.d/. Pods started on the hosts using multus CNI will have the main interface eth0 for intra-cluster communication and additional interfaces net0, net1, ... for the additionally configured networks in the CNI config. See test.yaml.sample for an example deployment using multiple interfaces.