controller, They are directly use fastpath (Data path). However, Note that each Flow Table entry has two timers:
Start your SDN journey today!
Emerging Technologies Group, Red Hat.
Objectives and Takeways
Session Objectives :
● High level overveiw of SDN Concepts and introduction of OpenvSwitch,
● Understand how to use OpenVswitch with KVM Virtualization
● Various OpenvSwitch Features with Demo
● SDN - OpenFlow
● SDN Controller (opendaylight project)
Key Takeways :
● Getting familiar with openvswitch and its cli utilities
● When to use which Feature of OpenvSwitch.
● Creating vlans for your TestBeds (KVM).
● Opendaylight controller and OpenFlow
What is SDN? Software Defined Network?
● Separation of the control from the forwarding plane.
● Software programmability for network elements.
● Centralized network control and management .
The control plane is where forwarding/routing decisions are made (Software Logic)
The data plane is where the data forwarding action takes place. (instructions to carry traffic
over hardware )
The SDN architecture decouples the network control and forwarding functions
enabling the network control to become directly programmable and the underlying
infrastructure to be abstracted for the applications and network service.
10000 foot view of SDN Approach
Notice how control and data planes are separated and how this separation helps applications to directly
controls network opening doors for innovation.
Network Infrastructure - This consists the network devices such as
routers and switches, Both physical as well as virtual.
Controller - This encompass the software based on a centralized
controller which could be on a server that talks to all the devices in
the network using open API’s, Like OpenFlow or OVMDB.
Applications : This encompasses the variety of application for which
the network exists. This includes voice, video, enterprise
applications, security appliances such as intrusion detection. These
application can to the controller using open API’s to give them what
they want. For example, Voice traffic may ask the controller to have it
treated with least latency while an enterprise backup server may tell
the controller to give it bandwith whenever it is available.
Limitations of Linux Bridge
Guest (VM) networking in KVM has traditionally been done using linux, It is simple to
configure and manage but is not originally designed for virtual networking and
therefore poses integration and management challenges.
An unmanaged software switch !
Robust, Does packet forwarding job very well.
Fast, simple to use, No complexity
Linux Bridge's “root cause problem”
It is Closed for Innovations!
What is OpenvSwitch?
OpenvSwitch is an open source openflow capable virtual switch. If you’re familiar
with VMware, think of it as an open source Distributed
● Flexible Control in user-space with comprehensive management tools
● Fast Datapath (forwarding plane) in kernel
● Adopted in base kernel, since version 3.3
● Open vSwitch was included since RHEL 6 .4 (only datapath, no Control).
● Any netdevice (physical/virtual) can be added as uplink port
The data path (ovs kernel module) uses netlink socket to interacts with vswitchd daemon that implements
and manages any number ovs switches on local system, SDN Controller interacts with vswitchd using
openflow protocol. The ovsdb-server maintains the switch table database and external clients can talk to
ovsdb-server using json rpc and json being the data format. ovsdb database currently contains around 13
tables and this database is persistent across restarts.
Starting with Fedora 16 the openvswitch user space tools and the
required kernel modules are included in the Fedora distribution.
# sudo dnf install openvswitch
openvswitch package contains all the required user space tools
including the ovsdb and series of command line utilities to
configure, monitor and manage open vswitch instances.
Following are the important configuration files of openvswitch.
RHEL 7 and later (Need openstack or openshift channel)
#rpm -qc openvswitch
Starting up the openvswitch service
The Openvswitch.service is comprised by two daemons. One is a database and
another is the switch itself.
#/bin/systemctl start openvswitch.service
#systemctl enable openvswitch.service
1. Kernel module gets loaded. module name is "openvswitch"
description: Open vSwitch switching datapath
2. Database is created. copied
Openvswitch command-line interfaces
#ovs-vsctl - This command is used to set up, maintain, and inspect various ovs switch
configurations. It provides a high level interface for Open vSwitch Database to query and apply
changes on runtime.
#ovs-ofctl and ovs-dpctl - These two commands are used for administering and monitoring
flow entries. You learned that OVS manages two kinds of flow.
OpenFlows - The flows managed at control plane.
Datapath - Kernel flow. A kind of cached version of the OpenFlow.
ovs-ofctl - speaks to OpenFlow module whereas ovs-dpctl - speaks to Kernel module. Following
two are most used options of each of these commands
ovs-appctl - This command offer a way to send commands to a running Open vSwitch and
gather information that is not directly exposed to ovs-ofctl command. This is swiss army knife of
Setting up your first openvswitch bridge
Multiple ways to Setup/Configure .
- The ovs-vsctl is the primary command to create, remove and administer openvswitch. The
operation performed using ovs-vsctl are persistent across system reboot.
[root@kvmHOST1 ~]# ovs-vsctl add-br vswitch001
[root@kvmHOST1 ~]# ovs-vsctl show
- Configuring a openvswitch bridge using network scripts.
[root@kvmHOST1]cat /etc/sysconfig/network-scripts/ifcfg- vswitch002
DEVICETYPE="ovs" Doc on ifcfg directives for openvswitch
Integrating KVM VMs and OVS
Edit VM XML Configuration
#virsh edit vm-name
Libvirt Network Pool
virsh net-define ovs-network.xml
Network vswitch-net defined from ovs-network.xml
# virsh net-start vswitch-net
Network vswitch-net started
A quick method!
#virt-xml vm_01 --edit --network
Feature 1 : Security / L2 Segregation
VLAN isolation enforces VLAN membership of a VM without the knowledge of the guest itself.
[root@kvmHOST1 ~]# ovs-vsctl set port fed1 tag=10
[root@kvmHOST1 ~]# ovs-vsctl set port fed2 tag=10
[root@kvmHOST1 ~]# ovs-vsctl set port fed3 tag=20
[root@kvmHOST1 ~]# ovs-vsctl set port fed4 tag=20
#ovs-vsctl set port fed1 trunks=20,30,40
#ovsdb-client monitor Port name,trunks --
The default, vlan_mode used is "access", Its native mechanism of the vlan. VLAN Tag
is added when packets enter a Access port, and stripped off when leave a access port.
Other vlan_mode are native−tagged, native−untagged and trunk
<portgroup name='novlan' default='yes'>
portgroup feature of libvirt provides a method of easily putting guest connections to the network into different
classes, with each class potentially having a different level/type of service.
Feature 2 : Overlay Networks
Overlay Networks are industry standard techniques designed achieve Network Virtualization.
Network Overlays such as Virtual eXtensible Local Area Network a(VXLAN) and Generic Routing
Encapsulation (GRE) achieve network virtualization by overlaying layer 2 networks over physical
layer 3 networks which enable network scalability and efficient use of current network
Openvswitch supports multiple tunneling protocols (GRE, VXLAN, STT, and Geneve, with
10.0.0.1 vxLan tunnle 10.0.0.2
192.168.1.10 underlay N/W 192.168.2.20
Feature 3 : QoS
The network QoS (quality of service) QoS refers to the ability of the network to handle
it traffic such that it meets the service needs of certain applications. It is often used as
a synonym for traffic control.
QoS Policing - Rate Limiting
Qos Shaping - Dedicated Queue
Input(inbound) and output(outbound) traffic
Feature 3 : QoS Configuration
Create a queue(q0) with required network
bandwidth. Here in this example I am limiting the
egress traffic bandwidth to 10 MBps.
Create a Queue
#ovs-vsctl --id=@q0 create queue other-config:min-
#ovs-vsctl create qos type=linux-htb
Add Qos to a port
#ovs-vsctl set port vent0 qos= 09f5b3c4-
To apply QoS on a VM to control its inbound traffic, Modify its
interface table to configure an ingress policing rule.
There are two rules to set:
ingress_policing_rate: The maximum rate
(in Kbps) that this VM should be allowed to send.
ingress_policing_burst: A parameter to the
policing algorithm to indicate the maximum amount
of data (in Kb) that this interface can send
beyond the policing rate.
ovs-vsctl set interface fed1 ingress_policing_rate=20000
ovs-vsctl set interface fed1 ingress_policing_burst=200
Feature 4 : Port Mirroring
OpenVswitch support port mirroring features out of the box, This feature is exactly
similar to the port mirroring capability available on the new generation physical
switches. With Port Mirroring Network administrator can get an insight on what kind of
traffic is flowing on the network.
#ovs-vsctl -- --id=@m create mirror name=M1 -- add bridge vswitch001 mirrors @m
$ovs-vsctl set port fed1 -- set mirror M1 select_src_port=@fed1 select_dst_port=@fed2
select_dst_port = Ports on which incoming packets are selected for
select_src_port = Ports on which outgoing packets are selected for mirroring.
select_all = Its boolean, when to true. every packet incoming or outgoing
on any port connected to bridge will be mirrored.
output_port = specify to which port we want to send this mirrored traffic
Feature 4 : Port Mirroring
SPAN (Switched Port Analyzer)
#ip link add dummy0 type dummy
# ovs-vsctl -- --id=@dummy0 get port dummy0 -- set mirror mymirror select_all=true
An SDN controller is an application in software-defined networking (SDN) that manages flow control to enable intelligent
Flow table is managed by a remote SDN controller, You can install or remove control
flows using the SDN controller connected to the bridge
Managing Openvswitch via a SDN Controller
By connecting an openvswitch to SDN controller, We get level of abstraction and
automation required to revolutionize networking.
#ovs-vsctl set-controller vswitch0 tcp:192.168.1.20:6633
More about openflow
OpenFlow allows creating powerful L2-L4 service insertion, A flow contains entries that
matches packets and apply actions that may include packet forwarding, packet
modification and others.
Basically rules are used to create expression, If expression matches, The defined Action is
Example : 1
cookie=0x0, duration=14.604s, table=0, n_packets=61, n_bytes=7418,
idle_timeout=10, hard_timeout=30,tcp, vlan_tci=0x0000, dl_src=52:54:00:CE:51:52,
dl_dst=52:54:00:CE:51:53, nw_src=10.0.0.1, nw_dst=10.0.0.2, nw_tos=0, tp_src=22,
Example 2 :
Where can I learn more about flow ? https://flowsim.flowgrammable.org/
- openvswitch Services Log files :
- VLOG : OpenvSwitch has a built-in logging mechanism called VLOG. The VLOG
facility expose deep internal information of various components. First Determine at
what level your problem is occurring, Is it Bonding problem?
ovs-appctl vlog/list | grep -i bond
- Verbosity of logging supported are (must be: emer, err, warn, info, or dbg),
#ovs-appctl vlog/set module[:facility[:level]]
ovsdb-client utility to query the ovs database
Tables : #ovsdb-client list-tables
Bridge, Queue,QoS,Port, Interface
# "ovsdb-client list-columns <table_name>" Prints columns in a particular table, There
are many columns in each table.
ovsdb-client monitor <table_name> <cloumn_name> --detach
- Some other handy tools
#ovsdb-tool showlog" to see data inserted into ovsdb, its openvswitch configuration
#ovs-vsctl --format=table --column=name,vlan_mode
"Show My Network State" is good utility to get graphically display of the virtual/physical
network topology inside a single host. The tools is available here for download,