Skip to content
/ sdn-lab2 Public

2025 Experiments for COMP550705 Software Defined Networks: Lab 2

2 stars 2 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

XJTU-NetVerify/sdn-lab2

BranchesTags

Repository files navigation

实验指导书（二）

[toc]

零、下载实验工程

我们使用UV来方便大家统一版本。如果你使用了我们的镜像，或者使用的是archlinux：

  1. 下载UV Python包管理器

    # （可选）更新软件列表
sudo pacman -Syy

# 安装实验所使用的Python包管理器
sudo pacman -S uv

    Fedora / RHEL

    sudo dnf install uv

    其他发行版

    curl -LsSf https://astral.sh/uv/install.sh | sh

  2. 获取实验Git仓库

    git clone https://github.com/XJTU-NetVerify/sdn-lab2.git
cd sdn-lab2

  3. 安装依赖和初始化

    uv sync

  4. 运行OS-Ken App

    uv run osken-manager simple_switch.py

    如果你不希望再反复输入uv run，你可以使用Python虚拟环境（virtualenv, venv）；此后你就可以直接运行：

    # 创建虚拟环境：
uv venv
uv sync

# 创建完虚拟环境后，每次进入新的shell你需要运行：
source .venv/bin/activate

# 此时和Conda一样，Shell prompt前面会有`(sdn-lab2)`提示，即说明你在`venv`虚拟环境中。
# 你可以直接运行osken-manager。
osken-manager simple_switch.py

# 如果要退出`venv`虚拟环境：
deactivate

一、OVS查看交换机流表

  • 查看交换机s1的流表
sudo ovs-ofctl dump-flows s1

image-20230319131750910

  • 只查看OpenFlow1.3的流表项
sudo ovs-ofctl -O OpenFlow13 dump-flows s1

image-20230319131820401

二、Wireshark对控制器抓包

启动WireShark，监听Loopback:lo端口。

image-20230319132413866

三、如何运行OS-Ken APP

Ryu现已不再维护；因此，我们在实验中选用从Ryu fork的，目前正在积极维护中的OS-Ken。其大部分内容和Ryu通用。

以运行OS-Ken提供的查看拓扑的APP为例：

  • mininet创建拓扑。
sudo mn --custom topo_2sw_2host.py --topo mytopo --controller remote
#remote表示不用`Mininet`自带的控制器，尝试使用`OS-Ken`等远端控制器。
  • 运行OS-Ken App。
uv run osken-manager simple_switch.py

四、OS-Ken编程示例：简单交换机

1. 代码

from os_ken.base import app_manager
from os_ken.controller import ofp_event
from os_ken.controller.handler import MAIN_DISPATCHER, CONFIG_DISPATCHER
from os_ken.controller.handler import set_ev_cls
from os_ken.ofproto import ofproto_v1_3

class L2Switch(app_manager.OSKenApp):
    OFP_VERSIONS = [ofproto_v1_3.OFP_VERSION]

    def __init__(self, *args, **kwargs):
        super(L2Switch, self).__init__(*args, **kwargs)
    
    def add_flow(self, datapath, priority, match, actions):
        dp = datapath
        ofp = dp.ofproto
        parser = dp.ofproto_parser

        inst = [parser.OFPInstructionActions(ofp.OFPIT_APPLY_ACTIONS, actions)]
        mod = parser.OFPFlowMod(datapath=dp, priority=priority, match=match, instructions=inst)
        dp.send_msg(mod)
 
	# add default flow table which sends packets to the controller
    @set_ev_cls(ofp_event.EventOFPSwitchFeatures, CONFIG_DISPATCHER)
    def switch_features_handler(self, ev):
        msg = ev.msg
        dp = msg.datapath
        ofp = dp.ofproto
        parser = dp.ofproto_parser

        match = parser.OFPMatch()
        actions = [parser.OFPActionOutput(ofp.OFPP_CONTROLLER, ofp.OFPCML_NO_BUFFER)]
        self.add_flow(dp, 0, match, actions)

	# handle packet_in message
    @set_ev_cls(ofp_event.EventOFPPacketIn, MAIN_DISPATCHER)
    def packet_in_handler(self, ev):
        msg = ev.msg
        dp = msg.datapath
        ofp = dp.ofproto
        parser = dp.ofproto_parser

        actions = [parser.OFPActionOutput(ofp.OFPP_FLOOD)]
        out = parser.OFPPacketOut(
            datapath=dp, buffer_id=msg.buffer_id, in_port=msg.match['in_port'],actions=actions, data=msg.data)
        dp.send_msg(out)

2. 运行

  • mininet创建树状拓扑
sudo mn --controller remote --mac --topo=tree,2,2
  • 将上述代码保存为simple_switch.py，启动控制器
uv run osken-manager simple_switch.py
  • 在mininet的CLI中启动wireshark抓取端口h4-eth0
mininet> h4 wireshark &
  • 在mininet中h1 ping h3，查看wireshark中关于h4的抓包情况

image-20230319131939977

image-20230319132013832

抓包结果显示该交换机存在验证缺陷：packet_in_handler函数会将数据包洪泛到交换机的所有端口，故h1和h3通讯时，h4也会收到所有的包。

五、实验任务

1. 实验任务一：自学习交换机

ARPANET_1969.png

1969年的ARPANET非常简单，仅由四个结点组成。假设每个结点都对应一个交换机，每个交换机都具有一个直连主机，你的任务是实现不同主机之间的正常通信。

前文给出的简单交换机洪泛数据包，虽然能初步实现主机间的通信，但会带来不必要的带宽消耗，并且会使通信内容泄露给第三者。因此，请你在简单交换机的基础上实现二层自学习交换机，避免数据包的洪泛。

问题说明

  • SDN自学习交换机的工作流程可以参考：

    1. 控制器为每个交换机维护一个mac-port映射表。
    2. 控制器收到packet_in消息后，解析其中携带的数据包。
    3. 控制器学习src_mac - in_port映射。
    4. 控制器查询dst_mac，如果未学习，则洪泛数据包；如果已学习，则向指定端口转发数据包(packet_out)，并向交换机下发流表项(flow_mod)，指导交换机转发同类型的数据包。

  • 网络拓扑为topo_1969_1.py，启动方式：

    sudo ./topo_1969_1.py

  • 可以不考虑交换机对数据包的缓存(no_buffer)。

代码框架

以下给出代码框架，只需补充关键的若干行实现即可：

from os_ken.base import app_manager
from os_ken.controller import ofp_event
from os_ken.controller.handler import MAIN_DISPATCHER, CONFIG_DISPATCHER
from os_ken.controller.handler import set_ev_cls
from os_ken.ofproto import ofproto_v1_3
from os_ken.lib.packet import packet
from os_ken.lib.packet import ethernet

class Switch(app_manager.OSKenApp):
    
    OFP_VERSIONS = [ofproto_v1_3.OFP_VERSION]
    def __init__(self, *args, **kwargs):
        super(Switch, self).__init__(*args, **kwargs)
        # maybe you need a global data structure to save the mapping
        
    def add_flow(self, datapath, priority, match, actions,idle_timeout=0,hard_timeout=0):
        dp = datapath
        ofp = dp.ofproto
        parser = dp.ofproto_parser
        inst = [parser.OFPInstructionActions(ofp.OFPIT_APPLY_ACTIONS, actions)]
        mod = parser.OFPFlowMod(datapath=dp, priority=priority,
                                idle_timeout=idle_timeout,
                                hard_timeout=hard_timeout,
        						match=match,instructions=inst)
        dp.send_msg(mod)
        
    @set_ev_cls(ofp_event.EventOFPSwitchFeatures, CONFIG_DISPATCHER)
    def switch_features_handler(self, ev):
        msg = ev.msg
        dp = msg.datapath
        ofp = dp.ofproto
        parser = dp.ofproto_parser
        match = parser.OFPMatch()
        actions = [parser.OFPActionOutput(ofp.OFPP_CONTROLLER,ofp.OFPCML_NO_BUFFER)]
        self.add_flow(dp, 0, match, actions)
        
    @set_ev_cls(ofp_event.EventOFPPacketIn, MAIN_DISPATCHER)
    def packet_in_handler(self, ev):
        msg = ev.msg
        dp = msg.datapath
        ofp = dp.ofproto
        parser = dp.ofproto_parser
        
        # the identity of switch
        dpid = dp.id
        # the port that receive the packet
        in_port = msg.match['in_port']
        pkt = packet.Packet(msg.data)
        eth_pkt = pkt.get_protocol(ethernet.ethernet)
        # get the mac
        dst = eth_pkt.dst
        src = eth_pkt.src
        # we can use the logger to print some useful information
        self.logger.info('packet: %s %s %s %s', dpid, src, dst, in_port)
        
        # You need to code here to avoid the direct flooding
        # Have fun!
        # :)

结果示例

UCLA ping UTAH，UCSB不再收到相关数据包：

image-20230321133954258

image-20230321134035864

2. 实验任务二：处理环路广播

ARPANET_loop

UCLA和UCSB通信频繁，两者间建立了一条直连链路。在新的拓扑topo_1969_2.py中运行自学习交换机，UCLA和UTAH之间无法正常通信。分析流表发现，源主机虽然只发了很少的几个数据包，但流表项却匹配了上千次；WireShark也截取到了数目异常大的相同报文。

image-20230321134114562

image-20230321134204284

这实际上是ARP广播数据包在环状拓扑中洪泛导致的，传统网络利用生成树协议解决这一问题。在SDN中，不必局限于生成树协议，可以通过多种新的策略解决这一问题。以下给出一种解决思路，请在自学习交换机的基础上完善代码，解决问题：

当序号为dpid的交换机从in_port第一次收到某个src_mac主机发出，询问dst_ip的广播ARP Request数据包时，控制器记录一个映射(dpid, src_mac, dst_ip)->in_port。下一次该交换机收到同一(src_mac, dst_ip)但in_port不同的ARP Request数据包时直接丢弃，否则洪泛。

代码框架

from os_ken.base import app_manager
from os_ken.controller import ofp_event
from os_ken.controller.handler import MAIN_DISPATCHER, CONFIG_DISPATCHER
from os_ken.controller.handler import set_ev_cls
from os_ken.ofproto import ofproto_v1_3
from os_ken.lib.packet import packet
from os_ken.lib.packet import ethernet
from os_ken.lib.packet import arp
from os_ken.lib.packet import ether_types

ETHERNET = ethernet.ethernet.__name__
ETHERNET_MULTICAST = "ff:ff:ff:ff:ff:ff"
ARP = arp.arp.__name__


class Switch_Dict(app_manager.OSKenApp):
    OFP_VERSIONS = [ofproto_v1_3.OFP_VERSION]

    def __init__(self, *args, **kwargs):
        super(Switch_Dict, self).__init__(*args, **kwargs)
        self.sw = {} #(dpid, src_mac, dst_ip)=>in_port, you may use it in mission 2
        # maybe you need a global data structure to save the mapping
        # just data structure in mission 1
        

    def add_flow(self, datapath, priority, match, actions, idle_timeout=0, hard_timeout=0):
        dp = datapath
        ofp = dp.ofproto
        parser = dp.ofproto_parser
        inst = [parser.OFPInstructionActions(ofp.OFPIT_APPLY_ACTIONS, actions)]
        mod = parser.OFPFlowMod(datapath=dp, priority=priority,
                                idle_timeout=idle_timeout,
                                hard_timeout=hard_timeout,
                                match=match, instructions=inst)
        dp.send_msg(mod)

    @set_ev_cls(ofp_event.EventOFPSwitchFeatures, CONFIG_DISPATCHER)
    def switch_features_handler(self, ev):
        msg = ev.msg
        dp = msg.datapath
        ofp = dp.ofproto
        parser = dp.ofproto_parser
        match = parser.OFPMatch()
        actions = [parser.OFPActionOutput(ofp.OFPP_CONTROLLER, ofp.OFPCML_NO_BUFFER)]
        self.add_flow(dp, 0, match, actions)

    @set_ev_cls(ofp_event.EventOFPPacketIn, MAIN_DISPATCHER)
    def packet_in_handler(self, ev):
        msg = ev.msg
        dp = msg.datapath
        ofp = dp.ofproto
        parser = dp.ofproto_parser

        # the identity of switch
        dpid = dp.id
        # the port that receive the packet
        in_port = msg.match['in_port']
        pkt = packet.Packet(msg.data)
        eth_pkt = pkt.get_protocol(ethernet.ethernet)
        if eth_pkt.ethertype == ether_types.ETH_TYPE_LLDP:
            return
        if eth_pkt.ethertype == ether_types.ETH_TYPE_IPV6:
            return
        # get the mac
        dst = eth_pkt.dst
        src = eth_pkt.src
        # get protocols
        header_list = dict((p.protocol_name, p) for p in pkt.protocols if type(p) != str)
        if dst == ETHERNET_MULTICAST and ARP in header_list:
            pass
        # you need to code here to avoid broadcast loop to finish mission 2
        
        # self-learning
        # you need to code here to avoid the direct flooding
        # having fun
        # :)
        # just code in mission 1

结果示例

解决ARP数据包在环状拓扑中的洪泛问题后，UCLA和UTAH之间可以ping通，并且流表项的匹配次数明显减少：

image-20230321134234564

3. 附加题

实验任务二只给出了一种参考方案，SDN中还有多种方案可供选择，请尝试设计实现一种新的策略解决环路广播问题。

附加题占分值较小，主要面向学有余力的同学，请结合自身实际决定是否选做。

4. 拓展资料

About

2025 Experiments for COMP550705 Software Defined Networks: Lab 2

Resources

Readme
Activity
Custom properties

Stars

2 stars

Watchers

0 watching

Forks

2 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages