kaposnick/k8s-open5gs

Code for the paper: Design and Validation of an Open Source Cloud Native Mobile Network

Design and Validation of an Open Source Cloud Native Mobile Network

The repository contains the code of the paper "Design and Validation of an Open Source Cloud Native Mobile Network" by N. Apostolakis, M. Gramaglia, P. Serrano. Please cite the paper if you plan to use it in your publication.

@ARTICLE{9877928,
  author={Apostolakis, Nikolaos and Gramaglia, Marco and Serrano, Pablo},
  journal={IEEE Communications Magazine}, 
  title={Design and Validation of an Open Source Cloud Native Mobile Network}, 
  year={2022},
  volume={60},
  number={11},
  pages={66-72},
  doi={10.1109/MCOM.003.2200195}}

Create the VM Network topology

• Latest Ubuntu 20.04 LTS version as VMs' base image
• Using KVM, virsh and virt-manager, create 3 VM domains; 1 for Master Node, 2 Worker Nodes. Give each one at least of 10G of disk space (this is important as while deploying the open5Gs pods, the node may suffer from disk pressure)
• Manage the VMs either using virsh console, or install an OpenSSH server
• Create the internal topology for control-plane and inter-pod communication. Also, create external interfaces to the worker nodes in order to provide S1-MME/N2 and S1-U/N3 connectivity towards the RAN

Create the Kubernetes Cluster

• Install Docker, kubelet, kubeadm in all the nodes
• Additionally, install kubectl in the master node
• Issue kubeadm init in the master node to create the Kubernetes Cluster. Use the argument --apiserver-advertise-address <interface-ip> to specify the IP address to be advertised to the worker nodes for the control plane communication.
• Issue kubeadm join in the worker nodes using the credentials provided by the master node.* Annotate the 2 Worker Nodes with mobile-core: cp for the node containing Control Plane CNFs and mobile-core: up for the node contanining User Plane CNFs.
  
• Setup the CNI: Flannel is going to be used as the container networking interface. Modify the --iface argument within the kube-flannel.yaml file in order to match the interface interconnecting the master node with the worker nodes and issue kubectl apply -f kube-flannel.yml in order to install the CNI in all of the worker nodes
• Make sure that the dockerd service in the worker nodes use the CNI for setting up the communication between the containers

Deploy the Open5Gs application

1. Modify the k8s-5g-core/values.yaml

The administrator should identify the externally accessible IP addresses exported by the 2 worker nodes and replace the

• mmeExternalAddress, webUiExternalAddress values with that of the C-Plane node.
• sgwuExternalAddress value with that of the U-Plane node.

2. Create the open5gs namespace

cd k8s-5g-core
kubectl create namespace open5gs

3. Deploy the application within the open5gs namespace

helm install open5gs -n open5gs .

4. Verify the state

$ kubectl get pods -n open5gs -o wide
NAME                              READY   STATUS    RESTARTS        AGE
mongo-0                           1/1     Running   0               3h40m
open5gs1-amf-deployment-0         1/1     Running   0               3h40m
open5gs1-ausf-deployment-0        1/1     Running   0               3h40m
open5gs1-bsf-deployment-0         1/1     Running   0               3h40m
open5gs1-nrf-deployment-0         1/1     Running   0               3h40m
open5gs1-nssf-deployment-0        1/1     Running   0               3h40m
open5gs1-pcf-deployment-0         1/1     Running   0               3h40m
open5gs1-pgwc-deployment-0        1/1     Running   0               3h40m
open5gs1-pgwu-deployment-0        1/1     Running   0               3h40m
open5gs1-udm-deployment-0         1/1     Running   0               3h40m
open5gs1-udr-deployment-0         1/1     Running   0               3h40m
open5gs1-webui-844ffb56d6-4spp2   1/1     Running   0               3h40m

The following steps are optional in case we want to have performance metrics collection from both the EPC/5GC and the RAN, using Prometheus.

Deploy the Prometheus application (optional)

Deploy the Prometheus stack within the kubernetes cluster. This will install the ServiceMonitor K8S resource that will search for pods with label app: srsenb within the open5gs namespace and set them as Prometheus targets.

cd k8s-open5gs
helm install -n monitoring prometheus -f ../prometheus-values.yaml prometheus-community/kube-prometheus-stack

Deploy the srsENB proxy pods (optional)

The srsENB process must listen on an cluster-accessible IP address and port. Then for every srsENB process that we want to monitor, deploy a helm chart issuing:

cd srsenb-chart
helm install <srsenbname> -n open5gs --set enb.ip=<enb-ip> --set enb.port=<enb-port> .

When the Prometheus's ServiceDiscovery discovers the proxy pods, the srsENB will start getting polled and the metrics will starting getting received.

Copyright and license

MIT License

Copyright (c) 2021 Nikolaos Apostolakis, IMDEA Networks Institute

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.