Network Layer & IP Addressing: Routing, IPV4/IPV6, Subnetting, NAT, ICMP

TAKEAWAYS:

• IPv4 vs. IPv6: IPv4’s 32-bit addresses are exhausted; IPv6’s 128-bit addresses solve this but require new protocols (e.g., ICMPv6).
• Subnetting: Divides a network into smaller subnets using CIDR notation (e.g., 192.168.1.0/24) to optimize routing and reduce broadcast traffic.
• Routing Protocols: RIP (distance-vector) and OSPF (link-state) determine the best path between networks; OSPF is faster for large networks.
• NAT: Translates private IP addresses (e.g., 192.168.x.x) to a single public IP, enabling millions of devices to share internet access (used by ISPs like NTC).
• ICMP: Handles error messages (e.g., "Destination Unreachable") and diagnostics (ping uses ICMP Echo Request/Reply).
• Congestion Control: TCP’s slow-start and AIMD algorithms prevent network overload (critical for apps like Pathao’s real-time ride-matching).

1. The Network Layer: Core Functions

The network layer (Layer 3 of the OSI model) is responsible for:

• Logical addressing: Assigning unique IP addresses to devices (e.g., 192.168.1.1).
• Routing: Forwarding packets between networks using routers.
• Congestion control: Managing traffic to avoid network collapse.
• Fragmentation/reassembly: Breaking large packets into smaller ones for transmission (IPv4 only).

Layer 7: Application
Layer 6: Presentation
Layer 5: Session
Layer 4: Transport (TCP/UDP)
Layer 3: Network (IP, Routing)
Layer 2: Data Link (MAC, Switches)
Layer 1: Physical (Cables, Signals)

Why it matters: Without the network layer, devices couldn’t communicate across different networks (e.g., your laptop to a Daraz server in Kathmandu).

2. IP Addressing: IPv4 and IPv6

IPv4 Addressing

• Format: 32-bit address divided into network and host parts (e.g., 192.168.1.1).
• Classes: Divided into Class A (0–127), Class B (128–191), Class C (192–223), Class D (multicast), and Class E (reserved).
• Subnetting: Splits a network into smaller subnets using a subnet mask (e.g., /24 for 255.255.255.0).

Example: A company with IP 10.0.0.0/8 can subnet it into /16 blocks:

• 10.0.0.0/16 → 10.0.0.0 to 10.0.255.255
• 10.1.0.0/16 → 10.1.0.0 to 10.1.255.255

Drawbacks of IPv4:

IPv6 Addressing

• Format: 128-bit address (e.g., 2001:0db8:85a3::8a2e:0370:7334).
• Advantages:
  • No NAT needed (plenty of addresses).
  • Built-in security (IPsec mandatory).
  • Simpler header (faster routing).
  • Autoconfiguration (devices assign their own IP).

Example: IPv6 in Nepal’s Ncell network enables seamless 4G/5G roaming without NAT translation.

IPv4 Header (20–60 bytes):
- Version (4 bits)
- IHL (Internet Header Length)
- Type of Service (ToS)
- Total Length (16 bits)
- Identification, Flags, Fragment Offset
- Time to Live (TTL)
- Protocol (TCP/UDP/ICMP)
- Header Checksum
- Source/Destination IP (32 bits each)

IPv6 Header (40 bytes):
- Version (4 bits)
- Traffic Class (8 bits)
- Flow Label (20 bits)
- Payload Length (16 bits)
- Next Header (8 bits)
- Hop Limit (8 bits)
- Source/Destination IP (128 bits each)

3. Subnetting and CIDR

Classless Inter-Domain Routing (CIDR) replaces fixed class boundaries with variable-length subnet masks (VLSM).

Steps to Subnet:

1. Determine required hosts per subnet (e.g., 30 hosts → need 5 bits for hosts: 2^5 – 2 = 30).
2. Calculate subnet mask:
   • Default Class C mask: /24 (255.255.255.0).
   • Need 5 host bits → borrow 3 bits from network → new mask: /27 (255.255.255.224).
3. List subnets:
   • First subnet: 192.168.1.0/27 (hosts: 192.168.1.1 to 192.168.1.30).
   • Second subnet: 192.168.1.32/27 (hosts: 192.168.1.33 to 192.168.1.62).

Real-World Example: Daraz’s warehouse network uses subnetting to isolate:

• 10.1.0.0/16 → Order processing servers.
• 10.2.0.0/16 → Inventory databases.
• 10.3.0.0/16 → Employee workstations.

192.168.1.0/27  → 192.168.1.0   – 192.168.1.30 (Subnet 1)
192.168.1.32/27 → 192.168.1.32 – 192.168.1.62 (Subnet 2)
192.168.1.64/27 → 192.168.1.64 – 192.168.1.94 (Subnet 3)
...
192.168.1.224/27→ 192.168.1.224–192.168.1.254 (Subnet 8)

4. Routing Protocols

Routers use routing protocols to share network topology info. Two key types:

RIP Example:

• Router A knows:
  • Network 10.0.0.0/8 is 1 hop away.
  • Network 192.168.1.0/24 is 2 hops away.
• If a link fails, RIP takes 180 seconds to update routes (slow for large networks).

OSPF Example:

• Nepal Telecom (NTC) uses OSPF to route traffic between Kathmandu, Pokhara, and Biratnagar.
• Routers flood Link-State Advertisements (LSAs) to all nodes, creating a shortest-path tree using Dijkstra’s algorithm.

5. Network Address Translation (NAT)

Problem: IPv4 addresses are scarce, but private networks (e.g., homes, offices) need many IPs. Solution: NAT translates private IPs (e.g., 192.168.1.5) to a single public IP (e.g., 203.123.45.6).

Types of NAT:

Example: Your home router uses PAT to let:

• 192.168.1.2:54321 (your laptop) → 203.123.45.6:12345 (public).
• 192.168.1.3:65432 (your phone) → 203.123.45.6:23456 (public).

Private IP:Port → Public IP:Port
192.168.1.2:54321 → 203.123.45.6:12345 (YouTube)
192.168.1.3:65432 → 203.123.45.6:23456 (WhatsApp)
192.168.1.4:7890  → 203.123.45.6:34567 (eSewa)

Why NAT is used by ISPs:

• NTC uses NAT to let thousands of users share a few public IPs.
• Khalti uses NAT to route payments securely between banks and merchants.

6. Internet Control Message Protocol (ICMP)

ICMP is used for error reporting and diagnostics (not for carrying user data).

Key ICMP Messages:

Example: Nepal’s NEPSE stock exchange uses ICMP to monitor server health:

• If a server is down, ICMP "Destination Unreachable" alerts admins.

sequenceDiagram
    participant A as Your PC
    participant B as Router
    participant C as Google DNS
    A->>B: ICMP Echo Request (ping)
    B->>C: ICMP Echo Request
    C->>B: ICMP Echo Reply
    B->>A: ICMP Echo Reply
    Note over A,B: Round-trip time (RTT) = 20ms

7. Congestion Control in Networks

Congestion occurs when too many packets flood the network, causing delays or packet loss.

TCP Congestion Control Mechanisms:

1. Slow Start: Exponentially increase window size until loss occurs.
2. AIMD (Additive Increase, Multiplicative Decrease):
   • Increase window by 1 MSS (Maximum Segment Size) per RTT.
   • If loss detected, halve the window.
3. Fast Retransmit: Resend lost packets without waiting for timeout.

Example: Pathao’s ride-matching system uses TCP congestion control to:

• Prioritize ride requests during peak hours (e.g., 7–9 PM).
• Avoid network overload when thousands of users request drivers simultaneously.

Time →
|       |       |       |       |
|-------|-------|-------|-------|
1      2      3      4      5
|       |       |       |
|-------|-------|-------|
Slow Start: 1 → 2 → 4 → 8 MSS
|-------|-------|-------|
AIMD: +1 per RTT until loss
|-------|-------|
Loss → Window / 2

In the Real World

1. eSewa Payments:

   • Uses IPv4 with NAT to route transactions between banks (e.g., NMB, Global IME) and merchant servers.
   • Routing: OSPF ensures low-latency payment processing across Nepal’s data centers.

2. Ncell 4G/5G Network:

   • IPv6 adoption enables seamless handover between towers without NAT.
   • ICMP monitors tower health; if a tower fails, ICMP alerts trigger failover.

3. Daraz Order Fulfillment:

   • Subnetting: Isolates warehouse zones (e.g., 10.1.0.0/16 for orders, 10.2.0.0/16 for inventory).
   • Congestion Control: TCP’s AIMD prevents order-processing delays during sales (e.g., Dashain).

Exam Tip

What Examiners Look For

1. IPv4/IPv6 Comparison: Always mention address size, NAT need, header size, and security.
2. Subnetting Calculations: Show step-by-step borrowing of bits and list subnets in order.
3. Routing Protocols: Differentiate RIP (slow, hop-count) vs. OSPF (fast, link-state).
4. NAT: Explain PAT vs. static NAT and why it’s used (address conservation).
5. ICMP: Relate to ping/traceroute and error messages.
6. Congestion Control: Draw the TCP window growth graph and explain AIMD.

Common Mistakes to Avoid

• Forgetting to exclude network/broadcast addresses in subnet calculations.
• Confusing MAC (Layer 2) vs. IP (Layer 3) addresses.
• Not labeling fields in packet headers (e.g., TTL, Protocol).
• Ignoring real-world examples (examiners love Ncell, eSewa, or Daraz cases).

High-Score Strategy

• Draw diagrams: Always include:
  • A subnetting table (with first/last usable IPs).
  • A routing table (next-hop, metric).
  • A NAT translation table.
• Use numbers: "RIP has a 15-hop limit; OSPF uses Dijkstra’s algorithm."
• Link to Nepal: "NTC uses OSPF for backbone routing; NAT in home routers saves public IPs."