IEC Reference Foundation Overview
This document provides a general description about the reference foundation of IEC. The Integrated Edge Cloud (IEC) will enable new functionalities and business models on the network edge. The benefits of running applications on the network edge are - Better latencies for end users - Less load on network since more data can be processed locally - Fully utilize the computation power of the edge devices
Currently, the chosen operating system(OS) is Ubuntu 16.04 and/or 18.04. The infrastructure orchestration of IEC is based on Kubernetes, which is a production-grade container orchestration with rich running eco-system. The current container network interface(CNI) solution chosen for Kubernetes is project Calico, which is a high performance, scalable, policy enabled and widely used container networking solution with rather easy installation and arm64 support. In the future, Contiv/VPP or OVN-Kubernetes would also be candidates for Kubernetes networking.
Networking and HW Prerequisites
For a virutal deploy, minimum hardware requirement is 1 baremetal server (either x86_64 or aarch64) for a deploy with 3 VMs on it.
For a baremetal deploy minimum hardware requirement is 3 baremetal servers.
Networking requirements - TBD
Methods of Installation
To address a large variety of setups, multiple methods of deployment should be supported. Deployment works both on x86_64 and aarch64 hw.
|Method||Pros (current state)||Cons (current state)||Prerequisites|
|Other installer solutions (e.g. Airship)|
Kubernetes Install for Ubuntu
Install Docker as Prerequisite
Please follow docker installation guide for Ubuntu arm64 to install Docker CE:
Please select the 18.06 version since it is the latest version kubelet supported.
Turn off all swap devices and files with: Disable swap on your machine
Install Kubernetes with Kubeadm
kubeadm helps you bootstrap a minimum viable Kubernetes cluster that conforms to best practices which is a preferred installation method for IEC currently. Now we choose v1.13.0 as a current stable version of Kubernetes for arm64. Usually the current host(edge server/gateway)'s management interface is chosen as the Kubeapi-server advertise address which is indicated here as
The common installation steps for both Kubernetes master and slave node are given as Linux shell scripts:
For host setup as Kubernetes master:
--pod-network-cidr is the subnet of the pod network and must not overlap with any of the host networks
(see https://kubernetes.io/docs/setup/independent/create-cluster-kubeadm/#pod-network for more details)
--apiserver-advertise-address is the IP on the master node to use for advertising the master’s IP
(see https://kubernetes.io/docs/setup/independent/create-cluster-kubeadm/#initializing-your-master for more details)
--service-cidr can be set up to be used for service VIPs and must not overlap with any of the host networks.
The default value is "10.96.0.0/12". If you change the default, make sure you update the clusterIP address
in the yaml file when installing etcd (in the steps below)
To start using your cluster, you need to run (as a regular user):
if you are the ``root`` user.
For hosts setup as Kubernetes slave:
which will skip ca-cert verification. After the `slave` joining the Kubernetes cluster, in the master node, you could check the cluster node with the command:
Install the Calico CNI Plugin to Kubernetes Cluster
Now we install a Calico network add-on so that Kubernetes pods can communicate with each other. The network must be deployed before any applications. Kubeadm only supports Container Networking Interface(CNI) based networks for which Calico has supported.
Install the Etcd Database
Please use the following command to install etcd database.
Install the RBAC Roles required for Calico
Install Calico to system
Firstly, we should get the configuration file from web site and modify the corresponding image from amd64 to arm64 version. Then, by using kubectl, the calico pod will be created.
Deploy the Calico using following command:
In calico.yaml file, there is an option "IP_AUTODETECTION_METHOD" about choosing network interface. The default value is "first-found" which means the first valid IP address (except local interface, docker bridge). So if the number of network-interface is more than 1 on your server, you should configure it depends on your networking environments. If it does not configure it properly, there are some error about calico-node pod: "BGP not established with X.X.X.X".
Remove the taints on master node
Verification for the Work of Kubernetes
Now we can verify the work of Kubernetes and Calico with Kubernets pod and service creation and accessing based on Nginx which is a widely used web server.
Firstly, create a file named nginx-app.yaml to describe a Pod and service by:
then test the Kubernetes working status with the script:
Helm Install on Arm64
Helm is a tool for managing Kubernetes charts. Charts are packages of pre-configured Kubernetes resources. The installation of Helm on arm64 is as follows:
We would like to provide a walk through shell script (described in the following chapter) to automate the installation of Kubernetes and Calico in the future. But this README is still useful for IEC developers and users.
For issues or anything on the reference foundation stack of IEC, you could contact:
Trevor Tao: firstname.lastname@example.org
Jingzhao Ni: email@example.com
Jianlin Lv: firstname.lastname@example.org
Alternative Methods of Installation
Script Based Installation
Akraino IEC repository now provides an automated method, based on sh scripts, that handles all above steps.
- 2 nodes (virtual machines or baremetal) with a preinstalled operating system (Ubuntu 16.04/18.04) and passwordless-sudo capable user on them (password-based login via SSH enabled);
The following snippet will automatically handle all steps described above in the previous chapter:
OPNFV Fuel installer can be leveraged to automate the IEC prerequisites setup (e.g. baremetal operating system provisioning for baremetal clusters), as well as the IEC installation itself.
- 1 jumpserver node with preinstalled operating system (Ubuntu 16.04/18.04 or CentOS7) - will also be used as a hypvervisor for the IEC VMs - for single hypervisor deployments;
- 1 jumpserver node with preinstalled operating system + 3 baremetal nodes for multiple hypervisor deployments;
Supported configurations include, but are not limited to:
- single hypervisor node running 3 VMs dedicated to IEC;
- 3 baremetal nodes dedicated IEC (K8 directly on baremetal);
- 3 baremetal nodes running a virtual control plane (each baremetal node has 1 VM dedicated to IEC);
Upstream Fuel patch is currently undergoing final stages of review and is expected to be merged soon.
Once the Fuel patch lands upstream, deploying IEC (including handling its prerequisites, like creating the required VMs on the hypervisor) can be done using (e.g. for an AArch64 single-hypervisor POD):
Heat Orchestration Templates
- Openstack Ocata or latest
- 2 or more compute nodes with enough RAM and disk (128 GB of RAM, 2 TB disk space)
- DPDK is optional but it is recommended
The scripts and templates can be found in the Akraino iec git repository:
More useful information can be found in the README in the same directory