mininet中的网络实验，数据包走的是内核协议栈。其好处就是实际网络协议栈的性能。缺点就是，要验证新的想法，需要修改内核协议栈。而且关于内核中的数据，比如tcp的拥塞窗口数据，可以参考[4-5]，能否用在mptcp上就不知道了。

1.试验步骤：

1安装mptcp协议栈，参考[1]

2安装mininet。

3运行柯拓补脚本[2]，multihomed-4h.py脚本，启动iperf软件，可以测量节点间的带宽。

4在运行iperf的时候，采用xterm启动了两个控制台，xterm h1 h2。但是要抓包，需要另启动一个h1的控制台。

5在xterm h1 启动后，在控制台输入抓包包命令：tcpdump -XX -n -i h1-eth0 -w /yourpath/mptcp.pcap

multihomed-4h.py

#!/usr/bin/python
from mininet.topo import Topo
from mininet.net import Mininet
from mininet.cli import CLI
from mininet.link import TCLink
import time
##https://serverfault.com/questions/417885/configure-gateway-for-two-nics-through-static-routeing
#    ____h2____
#   /          \
# h1           h3
#   \___h4_____/
#
max_queue_size = 100  
net = Mininet( cleanup=True )
h1 = net.addHost('h1',ip='10.0.1.1')
h2 = net.addHost('h2',ip='10.0.1.2')
h3 = net.addHost('h3',ip='10.0.2.2')
h4 = net.addHost('h4',ip='10.0.3.2')
c0 = net.addController( 'c0' )
net.addLink(h1,h2,intfName1='h1-eth0',intfName2='h2-eth0', cls=TCLink , bw=2, delay='10ms', max_queue_size=max_queue_size)
net.addLink(h2,h3,intfName1='h2-eth1',intfName2='h3-eth0', cls=TCLink , bw=2, delay='10ms', max_queue_size=max_queue_size)
net.addLink(h1,h4,intfName1='h1-eth1',intfName2='h4-eth0', cls=TCLink , bw=2, delay='30ms', max_queue_size=max_queue_size)
net.addLink(h4,h3,intfName1='h4-eth1',intfName2='h3-eth1', cls=TCLink , bw=2, delay='30ms', max_queue_size=max_queue_size)
net.build()
h1.setIP('10.0.1.1', intf='h1-eth0')
h1.cmd("ifconfig h1-eth0 10.0.1.1 netmask 255.255.255.0")

h1.setIP('10.0.3.1', intf='h1-eth1')
h1.cmd("ifconfig h1-eth1 10.0.3.1 netmask 255.255.255.0")

h1.cmd("ip route flush all proto static scope global")
h1.cmd("ip route add 10.0.1.1/24 dev h1-eth0 table 5000")
h1.cmd("ip route add default via 10.0.1.2 dev h1-eth0 table 5000")

h1.cmd("ip route add 10.0.3.1/24 dev h1-eth1 table 5001")
h1.cmd("ip route add default via 10.0.3.2 dev h1-eth1 table 5001")
h1.cmd("ip rule add from 10.0.1.1 table 5000")
h1.cmd("ip rule add from 10.0.3.1 table 5001")
h1.cmd("route add default gw 10.0.1.2  dev h1-eth0")

h2.setIP('10.0.1.2', intf='h2-eth0')
h2.setIP('10.0.2.1', intf='h2-eth1')
h2.cmd("ifconfig h2-eth0 10.0.1.2/24")
h2.cmd("ifconfig h2-eth1 10.0.2.1/24")
h2.cmd("ip route add 10.0.2.0/24 via 10.0.2.2")
h2.cmd("ip route add 10.0.1.0/24 via 10.0.1.1")
h2.cmd("echo 1 > /proc/sys/net/ipv4/ip_forward")


h4.setIP('10.0.3.2', intf='h4-eth0')
h4.setIP('10.0.4.1', intf='h4-eth1')
h4.cmd("ifconfig h4-eth0 10.0.3.2/24")
h4.cmd("ifconfig h4-eth1 10.0.4.1/24")
h4.cmd("ip route add 10.0.4.0  dev h4-eth1") #via 10.0.4.2
h4.cmd("ip route add 10.0.3.0 via 10.0.3.1")

h4.cmd("echo 1 > /proc/sys/net/ipv4/ip_forward")


h3.setIP('10.0.2.2', intf='h3-eth0')
h3.cmd("ifconfig h3-eth0 10.0.2.2 netmask 255.255.255.0")
h3.setIP('10.0.4.2', intf='h3-eth1')
h3.cmd("ifconfig h3-eth1 10.0.4.2 netmask 255.255.255.0")

h3.cmd("ip route flush all proto static scope global")
h3.cmd("ip route add 10.0.2.2/24 dev h3-eth0 table 5000")
h3.cmd("ip route add default via 10.0.2.1 dev h3-eth0 table 5000")

h3.cmd("ip route add 10.0.4.2/24 dev h3-eth1 table 5001")
h3.cmd("ip route add default via 10.0.4.1 dev h3-eth1 table 5001")
h3.cmd("ip rule add from 10.0.2.2 table 5000")
h3.cmd("ip rule add from 10.0.4.2 table 5001")

net.start()
time.sleep(1)
CLI(net)
net.stop()

2.The tips to run this script

3.And the packets captured results

以上是在wireshark中打开的mptcp.pcap的截图，可以看到，网络层使用的协议栈是mptcp，这对于上层应用是透明的。

4.reference

[1]mptcp install

[2]mptcp-topology

[3]mininet-topology

[4]tcpprobe, observe the TCP flow with kprobes

[5]PROPORTIONAL RATE REDUCTION FOR TCP