Network Plug-ins - Networking | Architecture

Overview

OpenShift Origin supports the Kubernetes Container Network Interface (CNI) as the interface between the OpenShift Origin and Kubernetes. Software defined network (SDN) plug-ins are a powerful and flexible way to match network capabilities to your networking needs. Additional plug-ins that support the CNI interface can be added as needed.

The following network plug-ins are currently supported by OpenShift Origin:

OpenShift SDN

OpenShift Origin deploys a software-defined networking (SDN) approach for connecting pods in an OpenShift Origin cluster. The OpenShift SDN connects all pods across all node hosts, providing a unified cluster network.

OpenShift SDN is installed and configured by default as part of the Ansible-based installation procedure. See the OpenShift SDN section for more information.

Flannel SDN

flannel is a virtual networking layer designed specifically for containers. OpenShift Origin can use it for networking containers instead of the default software-defined networking (SDN) components. This is useful if running OpenShift Origin within a cloud provider platform that also relies on SDN, such as OpenStack, and you want to avoid encapsulating packets twice through both platforms.

Architecture

OpenShift Origin runs flannel in host-gw mode, which maps routes from container to container. Each host within the network runs an agent called flanneld, which is responsible for:

Managing a unique subnet on each host
Distributing IP addresses to each container on its host
Mapping routes from one container to another, even if on different hosts

Each flanneld agent provides this information to a centralized etcd store so other agents on hosts can route packets to other containers within the flannel network.

The following diagram illustrates the architecture and data flow from one container to another using a flannel network:

Node 1 would contain the following routes:

default via 192.168.0.100 dev eth0 proto static metric 100
10.1.15.0/24 dev docker0 proto kernel scope link src 10.1.15.1
10.1.20.0/24 via 192.168.0.200 dev eth0

Node 2 would contain the following routes:

default via 192.168.0.200 dev eth0 proto static metric 100
10.1.20.0/24 dev docker0 proto kernel scope link src 10.1.20.1
10.1.15.0/24 via 192.168.0.100 dev eth0

Contiv SDN

Contiv is an open-source networking plug-in module for container infrastructure. Contiv provides an infrastructure for application-oriented network policies and support for a range of multiple networking modes. These include:

A configurable set of overlay networking modes.
Physical networking modes.
Support for industry-leading hardware.

OpenShift Origin can use Contiv for networking containers instead of the default OpenShift SDN.

Contiv configuration instructions are forthcoming.

Architecture

Each node within the cluster runs a Contiv agent called netplugin while the master hosts run the Contiv controller (called netmaster), along with supporting control plane components (such as etcd).

Together the components of Contiv (netmaster and netplugin) handle key networking functions for OpenShift Origin including:

Assigning IP addresses to each container pod on each cluster node.
Creating and managing multiple separate container network instances for different groups of containers.
Configuring the network forwarding layer components for layer two or layer three forwarding.
Configuring and enforcing a range of network policies.
Providing management interfaces (including both CLI and GUI) to configure and manage Contiv-specific features and configurations.
Providing an infrastructure for role-based controls that allow for multiple role-based network operations workflows.

Contiv uses the Container Network Interface (CNI) to interface with OpenShift Origin and Kubernetes. A key value store based on etcd is used to store Contiv-specific state information. This is in addition to and separate from the instance of etcd used by other components in the system, including OpenShift Origin itself.

Nuage SDN for OpenShift Origin

Nuage Networks' SDN solution delivers highly scalable, policy-based overlay networking for pods in an OpenShift Origin cluster. Nuage SDN can be installed and configured as a part of the Ansible-based installation procedure. See the Advanced Installation section for information on how to install and deploy OpenShift Origin with Nuage SDN.

Nuage Networks provides a highly scalable, policy-based SDN platform called Virtualized Services Platform (VSP). Nuage VSP uses an SDN Controller, along with the open source Open vSwitch for the data plane.

Nuage uses overlays to provide policy-based networking between OpenShift Origin and other environments consisting of VMs and bare metal servers. The platform’s real-time analytics engine enables visibility and security monitoring for OpenShift Origin applications.

Nuage VSP integrates with OpenShift Origin to allows business applications to be quickly turned up and updated by removing the network lag faced by DevOps teams.

Figure 1. Nuage VSP Integration with OpenShift Origin

There are two specific components responsible for the integration.

The nuage-openshift-monitor service, which runs as a separate service on the OpenShift Origin master node.
The vsp-openshift plug-in, which is invoked by the OpenShift Origin runtime on each of the nodes of the cluster.

Nuage Virtual Routing and Switching software (VRS) is based on open source Open vSwitch and is responsible for the datapath forwarding. The VRS runs on each node and gets policy configuration from the controller.

Nuage VSP Terminology

Figure 2. Nuage VSP Building Blocks

Domains: An organization contains one or more domains. A domain is a single "Layer 3" space. In standard networking terminology, a domain maps to a VRF instance.
Zones: Zones are defined under a domain. A zone does not map to anything on the network directly, but instead acts as an object with which policies are associated such that all endpoints in the zone adhere to the same set of policies.
Subnets: Subnets are defined under a zone. A subnet is a specific Layer 2 subnet within the domain instance. A subnet is unique and distinct within a domain, that is, subnets within a Domain are not allowed to overlap or to contain other subnets in accordance with the standard IP subnet definitions.
VPorts: A VPort is a new level in the domain hierarchy, intended to provide more granular configuration. In addition to containers and VMs, VPorts are also used to attach Host and Bridge Interfaces, which provide connectivity to Bare Metal servers, Appliances, and Legacy VLANs.
Policy Group: Policy Groups are collections of VPorts.

Mapping of Constructs

Many OpenShift Origin concepts have a direct mapping to Nuage VSP constructs:

Figure 3. Nuage VSP and OpenShift Origin mapping

A Nuage subnet is not mapped to an OpenShift Origin node, but a subnet for a particular project can span multiple nodes in OpenShift Origin.

A pod spawning in OpenShift Origin translates to a virtual port being created in VSP. The vsp-openshift plug-in interacts with the VRS and gets a policy for that virtual port from the VSD via the VSC. Policy Groups are supported to group multiple pods together that must have the same set of policies applied to them. Currently, pods can only be assigned to policy groups using the operations workflow where a policy group is created by the administrative user in VSD. The pod being a part of the policy group is specified by means of nuage.io/policy-group label in the specification of the pod.

Integration Components

Nuage VSP integrates with OpenShift Origin using two main components:

nuage-openshift-monitor
vsp-openshift plugin

nuage-openshift-monitor

nuage-openshift-monitor is a service that monitors the OpenShift Origin API server for creation of projects, services, users, user-groups, etc.

In case of a Highly Available (HA) OpenShift Origin cluster with multiple masters, nuage-openshift-monitor process runs on all the masters independently without any change in functionality.

For the developer workflow, nuage-openshift-monitor also auto-creates VSD objects by exercising the VSD REST API to map OpenShift Origin constructs to VSP constructs. Each cluster instance maps to a single domain in Nuage VSP. This allows a given enterprise to potentially have multiple cluster installations - one per domain instance for that Enterprise in Nuage. Each OpenShift Origin project is mapped to a zone in the domain of the cluster on the Nuage VSP. Whenever nuage-openshift-monitor sees an addition or deletion of the project, it instantiates a zone using the VSDK APIs corresponding to that project and allocates a block of subnet for that zone. Additionally, the nuage-openshift-monitor also creates a network macro group for this project. Likewise, whenever nuage-openshift-monitor sees an addition ordeletion of a service, it creates a network macro corresponding to the service IP and assigns that network macro to the network macro group for that project (user provided network macro group using labels is also supported) to enable communication to that service.

For the developer workflow, all pods that are created within the zone get IPs from that subnet pool. The subnet pool allocation and management is done by nuage-openshift-monitor based on a couple of plug-in specific parameters in the master-config file. However the actual IP address resolution and vport policy resolution is still done by VSD based on the domain/zone that gets instantiated when the project is created. If the initial subnet pool is exhausted, nuage-openshift-monitor carves out an additional subnet from the cluster CIDR to assign to a given project.

For the operations workflow, the users specify Nuage recognized labels on their application or pod specification to resolve the pods into specific user-defined zones and subnets. However, this cannot be used to resolve pods in the zones or subnets created via the developer workflow by nuage-openshift-monitor.

In the operations workflow, the administrator is responsible for pre-creating the VSD constructs to map the pods into a specific zone/subnet as well as allow communication between OpenShift entities (ACL rules, policy groups, network macros, and network macro groups). Detailed description of how to use Nuage labels is provided in the Nuage VSP Openshift Integration Guide.

vsp-openshift Plug-in

The vsp-openshift networking plug-in is called by the OpenShift Origin runtime on each OpenShift Origin node. It implements the network plug-in init and pod setup, teardown, and status hooks. The vsp-openshift plug-in is also responsible for allocating the IP address for the pods. In particular, it communicates with the VRS (the forwarding engine) and configures the IP information onto the pod.

Kuryr SDN for OpenShift Origin

Kuryr (or more specifically Kuryr-Kubernetes) is an SDN solution built using CNI and OpenStack Neutron. Its advantages include being able to use a wide range of Neutron SDN backends and providing interconnectivity between Kubernetes pods and OpenStack virtual machines (VMs).

Kuryr-Kubernetes and OpenShift Origin integration is primarily designed for OpenShift Origin clusters running on OpenStack VMs.

OpenStack Deployment Requirements

Kuryr SDN has some requirements regarding configuration of OpenStack it will be using. In particular:

Minimal service set is Keystone and Neutron.
A LBaaS service is required. It can be either Octavia (with all of its own dependencies) or deprecated Neutron-LBaaS v2.

kuryr-controller

kuryr-controller is a service responsible for watching OpenShift Origin API for new pods being spawned and creating Neutron resources for them. For example, when a pod gets created, kuryr-controller will notice that and call OpenStack Neutron to create a new port. Then, information about that port (or VIF) is saved into the pod’s annotations. kuryr-controller is also able to use precreated port pools for faster pod creation.

Currently, kuryr-controller must be run as a single service instance, so it is modeled in OpenShift Origin as Deployment with replicas=1. It requires access to the underlying OpenStack service APIs.

kuryr-cni

kuryr-cni is (currently) a dummy container that is responsible for installing and configuring Kuryr CNI plug-in on OpenShift Origin nodes. It should be running on every node, so it is modelled in OpenShift Origin as DaemonSet.

Kuryr CNI plug-in is called every time a new container that needs networking is spawned on the host. When that happens, the CNI plug-in starts watching OpenShift Origin API to wait for kuryr-controller providing VIF annotation. When annotation is noticed, CNI plug-in uses this info to plug the VIF to the containers.

F5 BIG-IP® Router Plug-in

A router is one way to get traffic into the cluster. The F5 BIG-IP® Router plug-in is one of the available router plugins.

The F5 router plug-in integrates with an existing F5 BIG-IP® system in your environment. F5 BIG-IP® version 11.4 or newer is required in order to have the F5 iControl REST API. The F5 router supports unsecured, edge terminated, re-encryption terminated, and passthrough terminated routes matching on HTTP vhost and request path.

The F5 router has feature parity with the HAProxy template router, and has additional features over the F5 BIG-IP® support in OpenShift v2. Compared with the routing-daemon used in earlier versions, the F5 router additionally supports:

path-based routing (using policy rules),
re-encryption (implemented using client and server SSL profiles)
passthrough of encrypted connections (implemented using an iRule that parses the SNI protocol and uses a data group that is maintained by the F5 router for the servername lookup).

Passthrough routes are a special case: path-based routing is technically impossible with passthrough routes because F5 BIG-IP® itself does not see the HTTP request, so it cannot examine the path. The same restriction applies to the template router; it is a technical limitation of passthrough encryption, not a technical limitation of OpenShift Origin.

Routing Traffic to Pods Through the SDN

Because F5 BIG-IP® is external to the OpenShift SDN, a cluster administrator must create a peer-to-peer tunnel between F5 BIG-IP® and a host that is on the SDN, typically an OpenShift Origin node host. This ramp node can be configured as unschedulable for pods so that it will not be doing anything except act as a gateway for the F5 BIG-IP® host. It is also possible to configure multiple such hosts and use the OpenShift Origin ipfailover feature for redundancy; the F5 BIG-IP® host would then need to be configured to use the ipfailover VIP for its tunnel’s remote endpoint.

F5 Integration Details

The operation of the F5 router is similar to that of the OpenShift Origin routing-daemon used in earlier versions. Both use REST API calls to:

create and delete pools,
add endpoints to and delete them from those pools, and
configure policy rules to route to pools based on vhost.

Both also use scp and ssh commands to upload custom TLS/SSL certificates to F5 BIG-IP®.

The F5 router configures pools and policy rules on virtual servers as follows:

When a user creates or deletes a route on OpenShift Origin, the router creates a pool to F5 BIG-IP® for the route (if no pool already exists) and adds a rule to, or deletes a rule from, the policy of the appropriate vserver: the HTTP vserver for non-TLS routes, or the HTTPS vserver for edge or re-encrypt routes. In the case of edge and re-encrypt routes, the router also uploads and configures the TLS certificate and key. The router supports host- and path-based routes.

Passthrough routes are a special case: to support those, it is necessary to write an iRule that parses the SNI ClientHello handshake record and looks up the servername in an F5 data-group. The router creates this iRule, associates the iRule with the vserver, and updates the F5 data-group as passthrough routes are created and deleted. Other than this implementation detail, passthrough routes work the same way as other routes.

When a user creates a service on OpenShift Origin, the router adds a pool to F5 BIG-IP® (if no pool already exists). As endpoints on that service are created and deleted, the router adds and removes corresponding pool members.
When a user deletes the route and all endpoints associated with a particular pool, the router deletes that pool.

F5 Native Integration

With native integration of F5 with OpenShift Origin, you do not need to configure a ramp node for F5 to be able to reach the pods on the overlay network as created by OpenShift SDN.

Also, only F5 BIG-IP® appliance version 12.x and above works with the native integration presented in this section. You also need sdn-services add-on license for the integration to work properly. For version 11.x, set up a ramp node.

Connection

The F5 appliance can connect to the OpenShift Origin cluster via an L3 connection. An L2 switch connectivity is not required between OpenShift Origin nodes. On the appliance, you can use multiple interfaces to manage the integration:

Management interface - Reaches the web console of the F5 appliance.
External interface - Configures the virtual servers for inbound web traffic.
Internal interface - Programs the appliance and reaches out to the pods.

An F5 controller pod has admin access to the appliance. The F5 image is launched within the OpenShift Origin cluster (scheduled on any node) that uses iControl REST APIs to program the virtual servers with policies, and configure the VxLAN device.

Data Flow: Packets to Pods

This section explains how the packets reach the pods, and vice versa. These actions are performed by the F5 controller pod and the F5 appliance, not the user.

When natively integrated, The F5 appliance reaches out to the pods directly using VxLAN encapsulation. This integration works only when OpenShift Origin is using openshift-sdn as the network plug-in. The openshift-sdn plug-in employs VxLAN encapsulation for the overlay network that it creates.

To make a successful data path between a pod and the F5 appliance:

F5 needs to encapsulate the VxLAN packet meant for the pods. This requires the sdn-services license add-on. A VxLAN device needs to be created and the pod overlay network needs to be routed through this device.
F5 needs to know the VTEP IP address of the pod, which is the IP address of the node where the pod is located.
F5 needs to know which source-ip to use for the overlay network when encapsulating the packets meant for the pods. This is known as the gateway address.
OpenShift Origin nodes need to know where the F5 gateway address is (the VTEP address for the return traffic). This needs to be the internal interface’s address. All nodes of the cluster must learn this automatically.
Since the overlay network is multi-tenant aware, F5 must use a VxLAN ID that is representative of an admin domain, ensuring that all tenants are reachable by the F5. Ensure that F5 encapsulates all packets with a vnid of 0 (the default vnid for the admin namespace in OpenShift Origin) by putting an annotation on the manually created hostsubnet - pod.network.openshift.io/fixed-vnid-host: 0.

A ghost hostsubnet is manually created as part of the setup, which fulfills the third and forth listed requirements. When the F5 controller pod is launched, this new ghost hostsubnet is provided so that the F5 appliance can be programmed suitably.

The term ghost hostsubnet is used because it suggests that a subnet has been given to a node of the cluster. However, in reality, it is not a real node of the cluster. It is hijacked by an external appliance.

The first requirement is fulfilled by the F5 controller pod once it is launched. The second requirement is also fulfilled by the F5 controller pod, but it is an ongoing process. For each new node that is added to the cluster, the controller pod creates an entry in the VxLAN device’s VTEP FDB. The controller pod needs access to the nodes resource in the cluster, which you can accomplish by giving the service account appropriate privileges. Use the following command:

$ oc adm policy add-cluster-role-to-user system:sdn-reader system:serviceaccount:default:router

Data Flow from the F5 Host

These actions are performed by the F5 controller pod and the F5 appliance, not the user.

The destination pod is identified by the F5 virtual server for a packet.
VxLAN dynamic FDB is looked up with pod’s IP address. If a MAC address is found, go to step 5.
Flood all entries in the VTEP FDB with ARP requests seeking the pod’s MAC address. ocated. An entry is made into the VxLAN dynamic FDB with the pod’s MAC address and the VTEP to be used as the value.
Encap an IP packet with VxLAN headers, where the MAC of the pod and the VTEP of the node is given as values from the VxLAN dynamic FDB.
Calculate the VTEP’s MAC address by sending out an ARP or checking the host’s neighbor cache.
Deliver the packet through the F5 host’s internal address.

Data Flow: Return Traffic to the F5 Host

These actions are performed by the F5 controller pod and the F5 appliance, not the user.

The pod sends back a packet with the destination as the F5 host’s VxLAN gateway address.
The openvswitch at the node determines that the VTEP for this packet is the F5 host’s internal interface address. This is learned from the ghost hostsubnet creation.
A VxLAN packet is sent out to the internal interface of the F5 host.

During the entire data flow, the VNID is pre-fixed to be 0 to bypass multi-tenancy.