Chapter 4: Network Layer (Control Plane)

🔹 1. Network Layer Overview: Core Ideas

✅ What is the Network Layer (Data Plane)?

• Responsible for routing packets from source to destination

• Encapsulates transport-layer (e.g., TCP/UDP) segments to form datagrams

• Includes:
• Forwarding: Process of receiving packet and sending it to the right output link
• Routing: Process of determining the path from source to destination
• Addressing: Assigning IP addresses to hosts and routers
• Packet Schedule Data Plane: Buffering, scheduling, queueing

⚠️ No reliability, flow, or congestion control — functions are primarily at transport layer

🔹 2. Router Architecture / Data Plane Components

✅ Key Parts of a Router

💡 Data Plane: Execution, Forwarding, Queuing, Scheduling — Network Layer Logic 💡 Control Plane: Design — Routing Algorithms, Forwarding Table, NAT/SDN Implications

🔹 3. Network Layer Protocols & Addressing

✅ IP: Internet Protocol

• most important network layer protocol

• Each IP datagram has:
• source IP address
• destination IP address
• header fields (e.g., fragment, TTL, checksum, options)

✅ IP is connectionless and stateless — best-effort, no guarantees

✅ Subnetting & CIDR

• CIDR (Classless Inter-Domain Routing): Allow routing by network prefixes

• Subnets:
• Subnet is group of IP addresses
• Same high-order bits (prefix) → common network portion
• Example:
• IP address 223.1.1.1/24 → subnet 223.1.1.0/24

💡 赇 prefix determines common routing; host part determines specific destinations

🔹 4. TCP/IP: A Working Example

• Transport (TCP / UDP) → Network (IP) → Link (Ethernet/WiFi) → Physical (copper/radio)

• Internet = layered system of packets + addressing

❗ This is the architecture of the Internet: flexible, scalable, layered protocol stacks

🔹 5. Subnets, Addressing, CIDR

✅ Subnet Example (10.0.0.0/24)

• Network portion: 10.0.0.0

• Host portion: 0.0.0.0 – 0.0.0.255

• You can have up to 254 IP addresses in this subnet

✅ CIDR: Classless

• CIDR notation: a.b.c.d/x, where x is prefix length

• Example: 200.23.16.0/20
• network portion = 20-bit
• host portion = 12 bits
• max 4096 IPs in this network

🔁 More specific route (more details) = more precise, better delegation

🔹 6. NAT: Network Address Translation

✅ NAT Overview

• NAT enables all devices in a local network to share a single public IPv4 address

• Uses a NAT table to map:
• LAN IP, port → WAN IP, port

• Often embedded in router, helps reduce address exhaustion

✅ NAT Translation Example

✅ Why NAT? → IPv4 Address exhaustion (only 4.3 billion address space)

🔹 7. IPv6: The Next Addressing Format

✅ Why IPv6?

• 128-bit address space → 3.4×10³⁸ IPs

• Simplified header (40 bytes) vs IPv4 (20+ bytes) → faster processing

• No checksum → sends to router

• No fragmentation → favors optimized routing at network layer

• Enables scalable routing, programmable flow control (e.g., P4, SDN)

✅ Address Format in IPv6

• Human-readable format: aaaa:b000:0000:0000:0000:0000:0000:0001

• Compression: aaaa::1 (leading, trailing zeros can be removed)

⚠️ IPv6 not yet fully adopted — but it's coming

✅ Why is IPv6 being used?

• Long time to deploy → 25 years!

• IPv4 running out → NAT ends up being the only solution

• Optimized packets: 40 bytes → Fast routing, less overhead

🔹 8. Packet Scheduling: Key Concepts

🧠 Key Insight: Scheduling and buffering determine network performance.

🔹 9. Congestion, Packet Loss, and Pack Buffering

✅ Understanding the "packet loss" problem

• Buffer overflow at router links → packets are dropped

• Multihop steps:
• Input queueing at one router
• Output queueing at next hop

• Both types can increase delay and cause loss

✅ Buffer Size Rule of Thumb

• Use RTT × link capacity (e.g., if R = 10 Gbps (10 billion bits/sec), RTT = 0.25 sec) → buffer ≈ 2.5 Gbit

🔹 10. Buffer Management: Real Life Implications

💡 Fixed QoS vs Dynamic BufferingSlot Allocation helps Slow, fast, or apply adaptive buffer sizes

🔹 11. Routing & Forwarding Fundamentals

✅ The 2 Key Processes

❗ Forwarding ≠ Routing ❗ Forwarding is local, routing is global

🔹 12. Generalized Forwarding: Match + Action

✅ Core Idea

• Use header fields to match incoming packet → then act

• Can be done at any layer (Link, Network, Transport)

🧠 Match + Action allows programmable network infrastructure — centralization + flexibility!

🔹 13. SDN: Software-Defined Networking

✅ What is SDN?

• Centralized control plane

• Row-based routing, flow-based forwarding

• Flow tables in switches → SDN controller manages them remotely

✅ SDN Advantages

• Easier control → policy across network easily enacted

• Modern flexibility → programmable network

• Future-oriented → matching OpenFlow, new P4, GFLOW tech

💡 SDN is the future → network programmability at scale

🔹 14. Routing Protocols: BGP, OSPF

✅ Need to know:

• BGP (Border Gateway Protocol) → inter-AS (networks of different ISP)

• OSPF (Open Shortest Path First) → intra-AS, uses link-state → Dijkstra

• RIP (Routing Information Protocol) → distance vector, not covered

📌 BGP: coordination between ISPs, IXPs 📌 OSPF: internal medium/high-speed routing amid many devices

🔹 15. NAT: Network Address Translation (cont'd)

✅ NAT simplifies addressing

• List of IP addr. inside your home → only one public address to the Internet

• Example: Home has 10 devices → 1 public IP

• NLRA (Network Layer Routing Architecture) is not affected by NAT

❗ NAT Issues

• End-to-End Principle: NAT violates this

• Contributes to network complexity

• Traversability: Hard for some Internet apps (e.g., WebRTC, VoIP)

• Status: Still widely used, particularly in home / ISP / mobile networks

🔹 16. Flow Tables: Match + Action Extensibility

✅ Fields in Flow Table

✅ Flow table = router logic = monetary traffic handling at scale!

🔹 17. Middleboxes and OpenFlow

✅ Middleboxes vs Routers

• NAT

• Firewall

• Cache

• Load Balancer

✅ OpenFlow for Middleboxes

✅ OpenFlow is NAT, Firewall, Cache, Load Balancer in one!

🔹 18. Buffering & Large Packets: Considerations

✅ Why Fragment?

• Mainly due to MTU (Maximum Transfer Unit) of the network interface

• If IP datagram exceeds MTU, it is fragmented into smaller datagrams

• Only reassembled at the destination → no intermediate reassembly

✅ Fragments: Identification

⚠️ Fragments are expensive → reassembly inefficient

🔹 19. OpenFlow and P4: Enabling Programmable Networks

✅ OpenFlow:

• Packet match → controller notification

• Can forward, drop, modify, or send to controller

• Integrated with SDN, Middleboxes, Firewalls, Caches

• Flow table allows:
• Per-packet decisions
• Rate shaping for traffic
• Direct link to central controller

✅ P4 (Programming Protocol-independent Data Plane)

• P4 is a language for programmable networking

• Allows you to program what actions the router takes

• Flow tables from OpenFlow can use P4 for instance-level control

✅ Why both? Let the hardware know network rules via P4 code → OpenFlow reapplies

🔹 20. Mid-Box: A Practical Overview

💡 Install iPerf + Wireshark → measure network bandwidth, simulate congestion

🔹 21. Addressing & Routing in Real Networks (Summary)

✅ Addressing & Routing Assembly

✅ Mandatory: Address manufacturing !

🔹 22. TCP/IP Protocol Suite: Architecture & Layering

• IP is the core of the network
• No guarantee for delivery, timing, or ordering

• Transport layer (TCP, UDP) manages reliability, flow, ordering

• Network layer deals with route selection, fragmentation, packet routing

• Link layer (Ethernet, WiFi, etc.) sends bit stream over physical connections

• Application layer (HTTP, SMTP, DNS) uses network layer (IP) to send packets to others

🧠 IP = network layer → data plane core of Internet

🔹 23. MAC vs IP: Routing Context

Overview:

• MAC address is 128-bit, used for link layer (e.g., Ethernet, WiFi)

• IP address is 32-bit (for IPv4), 128-bit (for IPv6) — used for network layer

• MAC addresses are used internally, while IP addresses are used externally

🔹 Key Transitions

• Link layer (MAC) → Network layer (IP) • Network layer (IP) → Transport (TCP / UDP) • Transport → Application (DNS, HTTP)

✅ Network layer can store packet in buffer, queue, schedule or drop

🔹 24. Exam Checklist (Chapter 4)

🔹 25. Introduction to P4 (Software-Defined Networks)

✅ Key Idea: P4 = Language for the network

• Complements OpenFlow → code the flow table

• P4 is language for the hardware → controller manages code

• Programmable datagram processing at network interface level → controls forwarding and queueing

✅ P4 is the main programming tool for SDN networks.

🔹 26. Final Notes on Chapter 4

🧠 This is the core of the network layer — buffering, addressing, routing, fragments, and flow control

💡 Key differentiator of the Internet: simple, consistent data-plane 💡 TCP uses TCP ACKs and duration-based timeouts for delivery 💡 IP handles Flow (responsible for packet management)

🧩 Move up to Chapter 5: Running Networked Applications 🔁 Build understanding of routing protocols with BGP, OSPF 🚀 Explore Internet layer with IP, NAT, IPv6 🧭 Traffic Engineering = Type of routing algorithm to maximize network efficiency

📚 Practice Tools for Chapter 4

• Wireshark: Capture TCP/IP packets, analyze header fields

• ipconfig, ifconfig: check current IP address, subnet, DHCP status

• traceroute, tracert: see network path, witness subnets and router transitions

• iperf: measure network bandwidth, simulate congestion

🧑‍🤝‍♂️ Pro Tip for Exams

Describe the network layer as:

• Abstraction between transport & link

• Focus on packets, not messages

• Many packet fields, some buffers, some queues, some scheduling

• Movement from input interface to output interface based on forwarding table

• IP is 32-bit, no guarantee, newer versions are 128-bit

Suggestаш: Given these questions, craft your answer around packet flows, NAT, CIDR, and IPv6, and tools like Wireshark to validate!

✅ Summary Table – Network Layer Overview

✅ Exam Must-Know Checklist (Chapter 4)

📌 Where to Go Next?

• Chapter 5: Running Networked Applications, Go-Back-N, TCP Flow & Congestion Control

• Chapter 6: Link-Layer, Ethernet, WiFi

• Chapter 7: Routing Protocols (BGP, OSPF), Flows, and IP address allocation by ISP

🧠 Your understanding of the data-plane is now solid Prepare for Extended Assignments that build on real-world packet processing!