| Project: | Container4NFV, https://wiki.opnfv.org/display/OpenRetriever/Container4NFV |
|---|---|
| Editors: | Xuan Jia (China Mobile), Gergely Csatari (Nokia) |
| Authors: | Container4NFV team |
| Abstract: | This document provides the users with top-down gap analysis regarding OpenRetriever feature requirements with OPNFV Installer, OpenStack Official Release and Kubernetes Official Release. |
Analyzis of the architecture options were moved to the Container4NFV wiki.
This section provides users with Container4NFV gap analysis regarding feature requirement with OPNFV Installer in Danube Official Release. The following table lists the use cases / feature requirements of container integrated functionality, and its gap analysis with OPNFV Installer in Danube Official Release. OPNFV installer should support them.
| Use Case / Requirement | Supported in Danube | Notes |
|---|---|---|
| Use Openstack Magnum to install container environment | No | Magnum is supported in Openstack Official Release, but it’s not supported in OPNFV Installer. Magnum is the place where container can be installed in OPNFV. |
| Use Openstack Ironic to supervise bare metal machine | No | Container could be installed in bare metal machine. Ironic provides bare metal machine, work with Magnum together to setup a container environment, be installed in OPNFV. |
| Use Openstack Kuryr to provide network for container | No | Container has its own network solution. Container needs to connect with virtual machines, and Kuryr which use Neutron provides network service is the best choice now. |
This section provides a gap analyzis between the targets of Container4NFV for release Euphrates (E) or later and the features provided by OpenStack in release Ocata. As the OPNFV and OpenStack releases tend to change over time this analyzis is planned to be countinously updated. During the analyzis all OpenStack projects considered.
(Editors note: Maybe we should define a scope of OpenStack projects which is considered. All OpenStack projects can mean anything.)
The following table lists the use cases / feature requirements of container integrated functionality, and its gap analysis with OpenStack.
| Use Case / Requirement | Related OpenStack project | Notes | Status |
|---|---|---|---|
| Manage container and virtual machine lifecycle with the same NB API | Zun or nova-docker driver | Magnum can deploy a Container Orchestration Engine (COE), but does not provide any lifecycle management operations to the containers deployed in the COE. Zun provides lifecycle management support for the containers deployed in the COE via Nova API, but not all COE API operations are supported. nova-docker driver provided support for container lifecycle management without a COE (and Magnum), but it was deprecated due to lack of community support. A fork of the original nova-docker driver is maintained by the Zun team to provide support for the sandbox containers. Note: Support for this is not targeted in OPNFV release E. | Open |
| Container private registry to store container images | Swift, Cinder, Glance, Glare | Container images need a storage backed from where the COE can serve the registry. This backend should be accessible and should be supported by the COE. As a workaround it is possible to install a registry backend to a VM , but it is more optimal to use the possible backends already available in OpenStack, like Swift, Cinder, Glance or Glare. | Open |
| Kuryr needs to support MACVLAN and IPVLAN | Kuryr | Using MACVLAN or IPVLAN could provide better network performance. It is planned for Ocata. | Open |
| Kuryr Kubernetes integration is needed | Kuryr | It is done in the frame of Container4NFV. | Targeted to OPNFV release E /OpenStack Ocata |
| HA support for Kuryr | Kuryr | Targeted to OPNFV release E /OpenStack Ocata | |
| HA support for Zun | Zun | Open |
This section provides users with Container4NFV gap analysis regarding feature requirement with Kubernetes Official Release. The following table lists the use cases / feature requirements of container integrated functionality, and its gap analysis with Kubernetes Official Release.
| Use Case / Requirement | Supported in v1.5 | Notes |
|---|---|---|
| Manage conainter and virtual machine in the same platform. | No | There are some ways how Kubernetes could manage VM-s:
|
| Kubernetes support multiple networks. | No | As VNF needs at least three interfaces. Management,control plane, data plane. CNI already supports multiple interfaces in the API definition. |
| Kubernetes support NAT-less connections to a container | No | SIP/SDP and SCTP are not working with NAT-ed networks |
| Kubernetes scheduling support CPU binding，NUMA features | No | The kubernetes schedular don’t support these features |
| DPDK need to support CNI | No | DPDK is the technology to accelerate the data plane. Container need support it, the same with virtual machine. |
| SR-IOV can support CNI (Optional) | No | SR-IOV could let container get high performance |
This quickstart shows you how to easily install a Kubernetes cluster on VMs running with Vagrant. You can find the four projects inside container4nfv/src/vagrant and their documentation: - kubeadm_basic: weave.rst - kubeadm_multus: multus.rst - kubeadm_ovsdpdk: ovs-dpdk.rst - kubeadm_virtlet: virtlet.rst
Vagrant is installed in Ubuntu 16.04 64bit. vagrant is to create kubernetes cluster using kubeadm. kubernetes installation by kubeadm can be refered to https://kubernetes.io/docs/getting-started-guides/kubeadm.
sudo apt-get install -y virtualbox wget –no-check-certificate https://releases.hashicorp.com/vagrant/1.8.7/vagrant_1.8.7_x86_64.deb sudo dpkg -i vagrant_1.8.7_x86_64.deb
git clone http://gerrit.opnfv.org/gerrit/container4nfv -b stable/euphrates cd container4nfv/src/vagrant/k8s_kubeadm/ vagrant up
vagrant ssh master -c “kubectl apply -f /vagrant/examples/virtio-user.yaml”
vagrant destroy -f
`
container4nfv/src/vagrant# ./setup_vagrant.sh
`
Consequently, we need to reboot to make libvirtd group effective.
To test all the projects inside vagrant/ just run the next script:
`
container4nfv/ci# ./deploy.sh
`
Using Joid to deploy Kubernetes in bare metal machine https://build.opnfv.org/ci/job/joid-k8-nosdn-nofeature-noha-baremetal-daily-euphrates/lastBuild/
Using Joid to deploy Kubernetes in bare metal machine with load balance enabled https://build.opnfv.org/ci/job/joid-k8-nosdn-lb-noha-baremetal-daily-euphrates/
measure network latency between containers in k8s using ping https://git.opnfv.org/yardstick/tree/tests/opnfv/test_cases/opnfv_yardstick_tc080.yaml
measure network latency between container and VM using ping https://git.opnfv.org/yardstick/tree/tests/opnfv/test_cases/opnfv_yardstick_tc081.yaml
This quickstart shows you how to easily install a Kubernetes cluster on VMs running with Vagrant. The installation uses a tool called kubeadm which is part of Kubernetes.
kubeadm assumes you have a set of machines (virtual or bare metal) that are up and running. In this way we can get a cluster with one master node and 2 workers (default). If you want to increase the number of workers nodes, please check the Vagrantfile inside the project.
[Multus](https://github.com/Intel-Corp/multus-cni) is a CNI proxy and arbiter of other CNI plugins.
With the help of Multus CNI plugin, multiple interfaces can be added at the same time when deploying a pod. Notably, Virtual Network Functions (VNFs) are typically requiring connectivity to multiple network interfaces.
The Multus CNI has the following features: - It is a contact between the container runtime and other plugins, and it doesn’t have any of its own net configuration, it calls other plugins like flannel/calico to do the real net conf. job. - Multus reuses the concept of invoking the delegates in flannel, it groups the multi plugins into delegates and invoke each other in sequential order, according to the JSON scheme in the cni configuration. - No. of plugins supported is dependent upon the number of delegates in the conf file. - Master plugin invokes “eth0” interface in the pod, rest of plugins(Mininon plugins eg: sriov,ipam) invoke interfaces as “net0”, “net1”.. “netn”. - The “masterplugin” is the only net conf option of Multus cni, it identifies the primary network. The default route will point to the primary network.
This quickstart shows you how to easily install a Kubernetes cluster on VMs running with Vagrant. The installation uses a tool called kubeadm which is part of Kubernetes.
kubeadm assumes you have a set of machines (virtual or bare metal) that are up and running. In this way we can get a cluster with one master node and 2 workers (default). If you want to increase the number of workers nodes, please check the Vagrantfile inside the kubeadm_basic/.
Nginx is a web server which can also be used as a reverse proxy, load balancer and HTTP cache.
This quickstart shows you how to easily install a Kubernetes cluster on VMs running with Vagrant. The installation uses a tool called kubeadm which is part of Kubernetes.
kubeadm assumes you have a set of machines (virtual or bare metal) that are up and running. In this way we can get a cluster with one master node and 2 workers (default). If you want to increase the number of workers nodes, please check the Vagrantfile inside the project.
Open vSwitch* with the Data Plane Development Kit [OvS-DPDK](http://openvswitch.org/) is a high performance, open source virtual switch.
Using DPDK with OVS gives us tremendous performance benefits. Similar to other DPDK-based applications, we see a huge increase in network packet throughput and much lower latencies.
CONTAINER4NFV setup a Kubernetes cluster on VMs running with Vagrant and kubeadm.
kubeadm assumes you have a set of machines (virtual or bare metal) that are up and running. In this way we can get a cluster with one master node and 2 workers (default). If you want to increase the number of workers nodes, please check the Vagrantfile inside the project.
Is Clearwater suitable for Network Functions Virtualization?
Network Functions Virtualization or NFV is, without any doubt, the hottest topic in the telco network space right now. It’s an approach to building telco networks that moves away from proprietary boxes wherever possible to use software components running on industry-standard virtualized IT infrastructures. Over time, many telcos expect to run all their network functions operating at Layer 2 and above in an NFV environment, including IMS. Since Clearwater was designed from the ground up to run in virtualized environments and take full advantage of the flexibility of the Cloud, it is extremely well suited for NFV. Almost all of the ongoing trials of Clearwater with major network operators are closely associated with NFV-related initiatives.
Clearwater follows IMS architectural principles and supports all of the key standardized interfaces expected of an IMS core network. But unlike traditional implementations of IMS, Clearwater was designed from the ground up for the Cloud. By incorporating design patterns and open source software components that have been proven in many global Web applications, Clearwater achieves an unprecedented combination of massive scalability and exceptional cost-effectiveness.
Clearwater provides SIP-based call control for voice and video communications and for SIP-based messaging applications. You can use Clearwater as a standalone solution for mass-market VoIP services, relying on its built-in set of basic calling features and standalone susbscriber database, or you can deploy Clearwater as an IMS core in conjunction with other elements such as Telephony Application Servers and a Home Subscriber Server.
Clearwater was designed from the ground up to be optimized for deployment in virtualized and cloud environments. It leans heavily on established design patterns for building and deploying massively scalable web applications, adapting these design patterns to fit the constraints of SIP and IMS. The Clearwater architecture therefore has some similarities to the traditional IMS architecture but is not identical.
This repository contains instructions and resources for deploying Metaswitch’s Clearwater project with Kubernetes.
If you need more information about Clearwater project please checkout our [documentation](https://github.com/opnfv/container4nfv/blob/master/docs/release/userguide/clearwater-project.rst) or the official repository.
The deployment exposes:
- the Ellis web UI on port 30080 for self-provisioning.
- STUN/TURN on port 3478 for media relay.
- SIP on port 5060 for service.
- SIP/WebSocket on port 5062 for service.
SIP devices can register with bono.:5060 and the Ellis provisioning interface can be accessed at port 30080.
CONTAINER4NFV uses setup_vagrant.sh to install all resource used by this repository.
container4nfv/src/vagrant# ./setup_vagrant.sh -b libvirt
Check clearwater/clearwater_setup.sh for details about k8s deployment.
container4nfv/src/vagrant/kubeadm_clearwater# ./deploy.sh
container4nfv/src/vagrant# ./cleanup.sh
It’s important to connect to Ellis to generate the SIP username, password and domain we will use with the SIP client. Use your <master ip addres> + port 30080 (k8s default port). If you are not which Ellis’s url is, please check inside your master node.
kubeadm_clearwater# vagrant ssh master
master@vagrant# ifconfig eth0 | grep "inet addr" | cut -d ':' -f 2 | cut -d ' ' -f 1
192.168.121.3
In your browser connect to <master_ip>:30080 (ex. 192.168.121.3:30080).
After that, signup and generate two users. The signup key is secret. Ellis will automatically allocate you a new number and display its password to you. Remember this password as it will only be displayed once. From now on, we will use <username> to refer to the SIP username (e.g. 6505551234) and <password> to refer to the password.
We’ll use both Twinkle and Blink SIP client. , since we are going to try this out inside a LAN network. This is, of course, only a local test inside a LAN network. Configure the clients may be a little bit trickie, so we add some screenshots:
