Introduction
Build a software-defined network using OpenFlow protocol, Mininet emulator, and ONOS controller with custom packet routing policies. This comprehensive guide covers everything from design through implementation, testing, and deployment.
Software-Defined Networking with OpenFlow
Build a software-defined network using OpenFlow protocol, Mininet emulator, and ONOS controller with custom packet routing policies.
DifficultyAdvanced
Duration4–5 weeks
Components10 items
Steps3 steps
Theory & Background
Traditional networking: control plane (routing decisions) and data plane (packet forwarding) embedded in each device — distributed intelligence. SDN: decouple control from forwarding. Centralized Controller programs all switches via OpenFlow protocol. Data plane: 'dumb' switches forward packets based on flow tables programmed by controller. Advantages: global network view enables better optimization, programmatic control (deploy new routing policies via code instantly), hardware independence (commodity switches with OpenFlow). Disadvantages: controller is single point of failure (redundancy needed), higher latency for first packet of new flow.
Advertisement
Components & Requirements
10 components required for this project.
| # | Component | Purpose | Qty |
|---|---|---|---|
| 1 | Mininet network emulator | Virtual SDN network simulation | x1 |
| 2 | ONOS SDN Controller | Centralized network controller | x1 |
| 3 | OpenFlow 1.3 switches (virtual) | Programmatic packet forwarding | x1 |
| 4 | Python (Ryu framework) | Custom SDN application development | x1 |
| 5 | Wireshark with OpenFlow dissector | OpenFlow message analysis | x1 |
| 6 | Ubuntu 20.04 (Mininet compatible) | Host OS | x1 |
| 7 | iperf3 | Network performance measurement | x1 |
| 8 | ONOS REST API | Network policy programming | x1 |
| 9 | OpenVSwitch (OVS) | OpenFlow-capable virtual switch | x1 |
| 10 | GNS3 (alternative) | Network simulation with real router images | x1 |
Step-by-Step Implementation
Follow these 3 steps carefully.
1
SDN Architecture Concepts
Traditional networking: control plane (routing decisions) and data plane (packet forwarding) embedded in each device — distributed intelligence. SDN: decouple control from forwarding. Centralized Controller programs all switches via OpenFlow protocol. Data plane: 'dumb' switches forward packets based on flow tables programmed by controller. Advantages: global network view enables better optimization, programmatic control (deploy new routing policies via code instantly), hardware independence (commodity switches with OpenFlow). Disadvantages: controller is single point of failure (redundancy needed), higher latency for first packet of new flow.
2
Mininet Network Emulation
Mininet creates a virtual network with real kernel networking stack. mn --topo tree,depth=3,fanout=3 creates a tree topology with 27 hosts and 13 switches. Custom topology in Python: class MyTopo(Topo): define hosts and switches, add links with bandwidth/delay parameters. Connect to Ryu controller: mn --controller=remote,ip=127.0.0.1,port=6653. Test connectivity: mininet> pingall. Measure throughput: mininet> iperf h1 h27. Mininet runs all hosts/switches on one Linux machine using network namespaces.
3
Ryu OpenFlow Application
Ryu is a Python-based SDN framework. OpenFlow application inherits from RyuApp. Event handlers: @set_ev_cls(ofp_event.EventOFPSwitchFeatures): called when switch connects, install initial flow rules. @set_ev_cls(ofp_event.EventOFPPacketIn): called when switch receives packet not matching any flow table entry. Handler: learn source MAC, look up destination MAC, if known install flow rule (avoiding future PacketIn events for this flow), otherwise flood.
Code & Implementation
Core code for sdn_controller.py:
sdn_controller.py
Python
from ryu.base import app_manager from ryu.controller import ofp_event from ryu.controller.handler import CONFIG_DISPATCHER, MAIN_DISPATCHER, set_ev_cls from ryu.ofproto import ofproto_v1_3 from ryu.lib.packet import packet, ethernet class L2Switch(app_manager.RyuApp): OFP_VERSIONS = [ofproto_v1_3.OFP_VERSION] def __init__(self, *args, **kwargs): super().__init__(*args, **kwargs) self.mac_to_port = {} # {dpid: {mac: port}} @set_ev_cls(ofp_event.EventOFPSwitchFeatures, CONFIG_DISPATCHER) def switch_connected(self, ev): """Install table-miss flow: send all unknown packets to controller""" dp = ev.msg.datapath ofp = dp.ofproto; parser = dp.ofproto_parser # Match: all packets. Action: send to controller match = parser.OFPMatch() actions = [parser.OFPActionOutput(ofp.OFPP_CONTROLLER)] inst = [parser.OFPInstructionActions(ofp.OFPIT_APPLY_ACTIONS, actions)] dp.send_msg(parser.OFPFlowMod(dp, priority=0, match=match, instructions=inst)) @set_ev_cls(ofp_event.EventOFPPacketIn, MAIN_DISPATCHER) def packet_in(self, ev): msg = ev.msg; dp = msg.datapath ofp = dp.ofproto; parser = dp.ofproto_parser pkt = packet.Packet(msg.data) eth = pkt.get_protocol(ethernet.ethernet) dpid = dp.id; in_port = msg.match['in_port'] # Learn source MAC self.mac_to_port.setdefault(dpid, {}) self.mac_to_port[dpid][eth.src] = in_port # Forward or flood out_port = self.mac_to_port[dpid].get(eth.dst, ofp.OFPP_FLOOD) # Install flow rule if destination known (avoid future PacketIn) if out_port != ofp.OFPP_FLOOD: match = parser.OFPMatch(in_port=in_port, eth_dst=eth.dst) actions = [parser.OFPActionOutput(out_port)] inst = [parser.OFPInstructionActions(ofp.OFPIT_APPLY_ACTIONS, actions)] dp.send_msg(parser.OFPFlowMod(dp, priority=1, match=match, instructions=inst)) # Forward this packet dp.send_msg(parser.OFPPacketOut(dp, msg.buffer_id, in_port, [parser.OFPActionOutput(out_port)], msg.data))Testing & Troubleshooting
Test Software-Defined Networking with OpenFlow by verifying each subsystem individually before full integration.
Troubleshooting Tips
Verify power voltages, check ground connections, use serial monitor for debug.
Real-World Applications
Data center fabric optimization
WAN traffic engineering
Cloud virtual networking
Campus network automation
Security enforcement with granular flow rules
Network slicing for 5G (SDN core)
Research and experimentation network
Network function virtualization (NFV) control
Extensions & Next Steps
- Implement QoS using DiffServ with OpenFlow marking
- Build a network-wide intrusion detection system using flow analytics
- Implement ECMP load balancing across multiple paths
- Create network topology visualization using ONOS northbound API
- Build intent-based networking where policies are expressed in high-level language
Interactive Playground
Coming SoonAn interactive simulator will be available here — simulate circuits and run code in-browser without hardware.
Frequently Asked Questions
Is SDN used in real production networks?
Yes, widely. Google's B4 WAN uses OpenFlow-based SDN to manage the entire backbone connecting their data centers — achieving near-100% link utilization vs 30–40% in traditional networks by globally optimizing traffic. Facebook (Meta) uses similar OpenDaylight-based SDN for data center networking. Microsoft Azure and AWS use SDN extensively for virtual networking (VPCs, security groups programmed as flow rules). SD-WAN (Software-Defined WAN) is a commercial application of SDN principles now used by enterprises for WAN cost optimization.
Advertisement