This tutorial is an introduction to how SDN-IP runs in practice. We'll go through starting up a simple emulated network, and we'll see how SDN-IP controls this network to move data from place to place.
If you haven't done so already, it's highly recommended that you go through the ONOS Tutorial first. This will give you some familiarity with the basic functionality of ONOS. In addition, you should read through the SDN-IP Architecture <LINK> document to get an overview of how SDN-IP works.
Hopefully you've already done the ONOS tutorial, so you already have the ONOS tutorial VM available. If not, check out the Setup your environment <LINK> section of the ONOS tutorial to get the VM ready.
We've prepared a simple emulated Mininet topology, which contains some OpenFlow switches to make up the SDN network. Connected around the edges of the SDN network are emulated routers. The routers run a piece of software called Quagga, which is an open-source routing suite. Note that it is not mandatory to use Quagga; any software/hardware capable of speaking BGP will do. In our case we run the BGP part of Quagga on them, to simulate external BGP routers belonging to other administrative domains. The goal of SDN-IP is to be able to talk BGP with these routers in order to exchange traffic between the different external ASes.
This figure shows the topology as observed by ONOS. We can see 6 blue OpenFlow switches, and 5 peripheral nodes with yellow icons.
Double-click the "SDN-IP Mininet" icon on the desktop to start up the network.
We can look at the configuration of the hosts in the Mininet terminal.
mininet> h1 ip addr show ... inet 192.168.1.1/24 brd 192.168.1.255 scope global h1-eth0 ... mininet> h2 ip addr show ... inet 192.168.2.1/24 brd 192.168.2.255 scope global h2-eth0 ...
Each host is in a different IP subnet. When SDN-IP is up and running, these hosts will be able to communicate with one another despite being in different networks. This is because the SDN network is able to route traffic based on BGP routes.
Also, double-click the "ONOS" icon on the desktop to start up the ONOS console. If you run the "devices" command, you should see the network has started up and connected to ONOS.
onos> devices id=of:00000000000000a1, available=true, role=MASTER, type=SWITCH, mfr=Nicira, Inc., hw=Open vSwitch, sw=2.3.0, serial=None, protocol=OF_13 id=of:00000000000000a2, available=true, role=MASTER, type=SWITCH, mfr=Nicira, Inc., hw=Open vSwitch, sw=2.3.0, serial=None, protocol=OF_13 id=of:00000000000000a3, available=true, role=MASTER, type=SWITCH, mfr=Nicira, Inc., hw=Open vSwitch, sw=2.3.0, serial=None, protocol=OF_13 id=of:00000000000000a4, available=true, role=MASTER, type=SWITCH, mfr=Nicira, Inc., hw=Open vSwitch, sw=2.3.0, serial=None, protocol=OF_13 id=of:00000000000000a5, available=true, role=MASTER, type=SWITCH, mfr=Nicira, Inc., hw=Open vSwitch, sw=2.3.0, serial=None, protocol=OF_13 id=of:00000000000000a6, available=true, role=MASTER, type=SWITCH, mfr=Nicira, Inc., hw=Open vSwitch, sw=2.3.0, serial=None, protocol=OF_13
If you try and ping between any two hosts right now, you'll notice nothing is working.
mininet> h1 ping h2 PING 192.168.2.1 (192.168.2.1) 56(84) bytes of data. From 192.168.1.254 icmp_seq=1 Destination Net Unreachable From 192.168.1.254 icmp_seq=2 Destination Net Unreachable From 192.168.1.254 icmp_seq=3 Destination Net Unreachable
Even though ONOS is running and connected to switches, there are no applications loaded so there is nothing to tell ONOS how to control the network. We can also use the summary command to verify there are no flows or intents in the network.
onos> summary node=127.0.0.1, version=1.0.0.SNAPSHOT nodes=1, devices=6, links=14, hosts=5, clusters=1, paths=46, flows=0, intents=0
First we need to install some helper applications that SDN-IP relies on. These features let ONOS read in various configuration files and respond to ARP requests on behalf of hosts.
onos> feature:install onos-app-config onos> feature:install onos-app-proxyarp
Now lets install the SDN-IP application so we can get some traffic flowing between our networks.
onos> feature:install onos-app-sdnip
A lot happens as soon as we install the SDN-IP application. The first thing it does is install point-to-point intents to allow the external BGP peers to communicate with our internal BGP speaker. This allows the external BGP routers to relay the routes they are capable of forwarding through to SDN-IP.
We can see the routes that SDN-IP has learnt with the "routes" command.
onos> routes prefix=192.168.1.0/24, nexthop=10.0.1.1 prefix=192.168.2.0/24, nexthop=10.0.2.1 prefix=192.168.3.0/24, nexthop=10.0.3.1
Don't worry if you don't see all of the routes straight away - sometimes it takes a minute or so for the BGP sessions to establish and advertise the routes to ONOS.
Now that ONOS has learnt some routes, it has programmed those routes into the switches using the intent API. If we look at the intent summary, we can see the different intents that SDN-IP is using.
onos> intents -s Connectivity total= 27 installed= 27 Connectivity withdrawn= 0 failed= 0 Connectivity submitted= 0 compiling= 0 Connectivity installing= 0 recompiling= 0 Connectivity withdrawing= 0 PointToPoint total= 24 installed= 24 PointToPoint withdrawn= 0 failed= 0 PointToPoint submitted= 0 compiling= 0 PointToPoint installing= 0 recompiling= 0 PointToPoint withdrawing= 0 MultiPointToSinglePoint total= 3 installed= 3 MultiPointToSinglePoint withdrawn= 0 failed= 0 MultiPointToSinglePoint submitted= 0 compiling= 0 MultiPointToSinglePoint installing= 0 recompiling= 0 MultiPointToSinglePoint withdrawing= 0 SinglePointToMultiPoint total= 0 installed= 0 SinglePointToMultiPoint withdrawn= 0 failed= 0 SinglePointToMultiPoint submitted= 0 compiling= 0 SinglePointToMultiPoint installing= 0 recompiling= 0 SinglePointToMultiPoint withdrawing= 0 Path total= 0 installed= 0 Path withdrawn= 0 failed= 0 Path submitted= 0 compiling= 0 Path installing= 0 recompiling= 0 Path withdrawing= 0 LinkCollection total= 0 installed= 0 LinkCollection withdrawn= 0 failed= 0 LinkCollection submitted= 0 compiling= 0 LinkCollection installing= 0 recompiling= 0 LinkCollection withdrawing= 0 UnknownType total= 0 installed= 0 UnknownType withdrawn= 0 failed= 0 UnknownType submitted= 0 compiling= 0 UnknownType installing= 0 recompiling= 0 UnknownType withdrawing= 0 All total= 27 installed= 27 All withdrawn= 0 failed= 0 All submitted= 0 compiling= 0 All installing= 0 recompiling= 0 All withdrawing= 0
We see a total of 27 intents. The 24 PointToPointIntents are simple end-to-end flows which allow the external BGP routers to communicate with our internal BGP speaker. The three MultiPointToSinglePoint intents are the forwarding rules for the routes that we've learnt through BGP. Each route is translated into one MultiPointToSinglePoint intent which matches the traffic for that route at the ingress ports of the network, and forwards it along to the router who advertised the route to us. This is how we use routing information learnt from BGP to enable traffic to transit our network on these routes.
Now that the intents are installed, we can ping through the network. Go back to the Mininet console and try ping between a pair of hosts.
mininet> h1 ping h2 PING 192.168.2.1 (192.168.2.1) 56(84) bytes of data. 64 bytes from 192.168.2.1: icmp_seq=1 ttl=62 time=0.693 ms 64 bytes from 192.168.2.1: icmp_seq=2 ttl=62 time=0.139 ms 64 bytes from 192.168.2.1: icmp_seq=3 ttl=62 time=0.149 ms
The ping should succeed. You can try pinging between some of the other hosts such as h1, h2 and h3. We can't ping h4 yet, but we'll address that in the next section.
Now that we've got a the system up and running, let's see what happens when there's a change in the BGP routes. We're going to make one of the external routers advertise a new route, which will allow us to talk to a new host. Right now r4 is not advertising any routes, and so we can't talk to h4. Let's verify this by trying to ping h4.
mininet> h1 ping h4 PING 192.168.4.1 (192.168.4.1) 56(84) bytes of data. From 192.168.1.254 icmp_seq=1 Destination Net Unreachable From 192.168.1.254 icmp_seq=2 Destination Net Unreachable From 192.168.1.254 icmp_seq=3 Destination Net Unreachable From 192.168.1.254 icmp_seq=4 Destination Net Unreachable
Yep, looks like we can't reach h4 yet.
To make r4 advertise a new route, we have to change the configuration of the BGP router. In our case, the BGP router is a Quagga process, so we'll connect to the Quagga CLI and configure r4 to advertise a new route. (The Quagga CLI is complex and includes lots of options, but considering this is not a Quagga tutorial we don't have the time to explain too much about Quagga. If you're interested, there's material online that will help you understand Quagga).
First, from the Mininet CLI we can start up an xterm so we can connect to the Quagga CLI.
mininet> xterm r4
A little xterm window will pop up. The next few commands will by typed into this xterm window.
We can use telnet to connect to the Quagga BGP CLI.
root@ubuntu:~# telnet localhost 2605 Trying 127.0.0.1... Connected to localhost. Escape character is '^]'. Hello, this is Quagga (version 0.99.23). Copyright 1996-2005 Kunihiro Ishiguro, et al. User Access Verification Password:
It will prompt for a password, which is sdnip. Once the password is entered, we'll drop into a prompt
Now we can begin to configure the router to advertise a new network.
r4> enable r4# configure terminal r4(config)# router bgp 65004 r4(config-router)# network 192.168.4.0/24 r4(config-router)# exit r4(config)# exit r4#
Now our external router r4 has advertised a new route to our SDN network. We're done with the xterm window, so you can close it. Let's go back to our ONOS terminal and see if ONOS has received the new route.
onos> routes prefix=192.168.1.0/24, nexthop=10.0.1.1 prefix=192.168.2.0/24, nexthop=10.0.2.1 prefix=192.168.3.0/24, nexthop=10.0.3.1 prefix=192.168.4.0/24, nexthop=10.0.4.1
We see the new route to 192.168.4.0/24 has appeared in the list. Also, when SDN-IP received the route it installed a new MultiPointToSinglePoint intent into the network.
onos> intents -s ... MultiPointToSinglePoint total= 4 installed= 4 MultiPointToSinglePoint withdrawn= 0 failed= 0 MultiPointToSinglePoint submitted= 0 compiling= 0 MultiPointToSinglePoint installing= 0 recompiling= 0 MultiPointToSinglePoint withdrawing= 0 ...
Our number of MultiPointToSinglePoint intents has increased to 4. Now let's see if we can ping to our new network.
mininet> h1 ping h4 PING 192.168.4.1 (192.168.4.1) 56(84) bytes of data. 64 bytes from 192.168.4.1: icmp_seq=1 ttl=62 time=0.595 ms 64 bytes from 192.168.4.1: icmp_seq=2 ttl=62 time=0.182 ms 64 bytes from 192.168.4.1: icmp_seq=3 ttl=62 time=0.164 ms ...
Great! Now we can ping to h4 which is in the network we just received through BGP. This shows that whenever the routes learnt through BGP are updated, SDN-IP reacts to the update and programs the dataplane accordingly.
Thanks for completing the SDN-IP tutorial. Now that you've had a glimpse of how SDN-IP works, feel free to poke around and explore the system further. If you're looking to run SDN-IP in your own network, head over to the Users Guild <LINK> to learn more about configuring and deploying the application.
In addition, the scripts used to create the network are in the ONOS repository. The Mininet topology is in
tools/test/topos/sdnip/tutorial.py. The router configurations are in
tools/test/topos/sdnip/configs/. Feel free to take a look and see how the network is put together if you're interested.