IPv4 Header & Routing: Packets, Fields, Algorithms & Real-World Paths

TAKEAWAYS:

• IPv4 headers carry 13 fields (e.g., version, TTL, protocol) that routers use to forward packets, with checksum ensuring header integrity.
• TTL (Time To Live) decrements at each hop; when it reaches 0, the packet is discarded (preventing infinite loops).
• Static routing uses manually configured tables (e.g., NTC’s backbone links), while dynamic routing (RIP, OSPF) adapts to topology changes (e.g., Pathao’s ride-matching servers).
• Subnetting divides a Class C network (e.g., 192.168.1.0/24) into smaller subnets (e.g., /28) to optimize host allocation and reduce broadcast traffic.
• Distance-vector (RIP) routers share full routing tables periodically, while link-state (OSPF) routers flood topology maps for faster convergence.
• Routing loops are prevented by split horizon, poison reverse, and hold-down timers in distance-vector protocols.

---

IPv4 Header: The 13 Fields That Steer Packets

Every IPv4 packet carries a 20-byte header (minimum) with 13 critical fields. Routers inspect these fields to forward packets correctly. Below is the header format with field sizes and purposes:

+-------------------------------+-------------------------------+
| Version (4 bits) | Header Length (4 bits) | Type of Service (8 bits) |
+-------------------------------+-------------------------------+
| Total Length (16 bits)        | Identification (16 bits)   |
+-------------------------------+-------------------------------+
| Flags (3 bits) | Fragment Offset (13 bits) |
+-------------------------------+-------------------------------+
| Time To Live (TTL, 8 bits)    | Protocol (8 bits)            |
+-------------------------------+-------------------------------+
| Header Checksum (16 bits)     | Source IP (32 bits)          |
+-------------------------------+-------------------------------+
| Destination IP (32 bits)      | Options (variable)           |
+-------------------------------+-------------------------------+
| Padding (variable)            |                             |
+-------------------------------+-------------------------------+

Key Fields Explained:

1. Version (4 bits)

   • Always 4 for IPv4 (vs. 6 for IPv6).
   • Why? Ensures receivers know how to parse the header.

2. Header Length (4 bits)

   • Specifies header size in 32-bit words (minimum = 5 words = 20 bytes).
   • Example: If options are added, this field increases (e.g., 6 for 24 bytes).

3. Type of Service (ToS, 8 bits)

   • Prioritizes packets (e.g., 00010000 for low delay, used by VoIP apps like Pathao’s call routing).
   • Deprecated in favor of DSCP (Differentiated Services Code Point) in modern networks.

4. Total Length (16 bits)

   • Maximum packet size: 65,535 bytes (header + data).
   • Example: A DNS query (UDP) might use 512 bytes; a large file transfer (TCP) could use 1,500 bytes (MTU of Ethernet).

5. Identification, Flags, Fragment Offset

   • Used for fragmentation when a packet exceeds the MTU (e.g., 1,500 bytes for Ethernet).
   • Real-world: Ncell’s core network fragments packets if they exceed the MTU of a satellite link.

6. TTL (Time To Live, 8 bits)

   • Starts at 64 (common default) or 128 (for long paths).
   • Decrements by 1 at each router. If TTL = 0, the packet is dropped, and an ICMP "Time Exceeded" message is sent.
   • Example: Tracing a route to google.com uses TTL to map the path:

     $ traceroute google.com
     1  192.168.1.1 (TTL=63)
     2  10.0.0.1 (TTL=62)
     3  203.123.45.6 (TTL=61)
     

7. Protocol (8 bits)

   • Tells the destination which upper-layer protocol to use:
     • 1 = ICMP (ping)
     • 6 = TCP (HTTPS, SSH)
     • 17 = UDP (DNS, VoIP)
   • Example: An HTTP request (TCP) has Protocol=6; a DNS query (UDP) has Protocol=17.

8. Header Checksum (16 bits)

   • Ensures header integrity (not data). Recalculated at each hop.
   • How? Uses a 16-bit one’s complement sum of the header.
   • Example: If a bit flips in the header (e.g., due to noise), the checksum fails, and the packet is discarded.

9. Source/Destination IP (32 bits)

   • Identifies endpoints. Routers use the destination IP to look up the next hop in their routing table.

10. Options (Variable, up to 40 bytes)

    • Rarely used today. Historically included:
      • Security (for military networks)
      • Record Route (debugging)
      • Timestamp (performance analysis)

---

How Routers Use the IPv4 Header

Routers perform three key actions using the header:

1. Check TTL: Decrement by 1. If TTL = 0, drop and send ICMP error.
2. Recalculate Checksum: Due to TTL decrement, the header changes, so the checksum must be updated.
3. Lookup Routing Table: Match the destination IP to find the next hop.

IMAGE: Cisco Router Interior | A real router’s routing table interface

Routing Table Entry:
Destination     Mask          Gateway     Interface  Metric
192.168.1.0/24  255.255.255.0  0.0.0.0      eth0       0
0.0.0.0/0       0.0.0.0       192.168.1.1  eth0       1

---

Routing Algorithms: How Networks Find the Best Path

Routing algorithms determine the optimal path for packets. They are classified into:

1. Static Routing
2. Dynamic Routing (Distance-Vector, Link-State, Hybrid)

Comparison Table: Static vs. Dynamic Routing

Real-World Example: Ncell’s Core Network

• Static Routing: Used for backbone links between major cities (e.g., Kathmandu to Pokhara). Admins manually configure routes to ensure low latency.
• Dynamic Routing (OSPF): Used within regional networks to reroute traffic if a tower fails (e.g., during a landslide).

---

Distance-Vector vs. Link-State Routing

1. Distance-Vector Routing (RIP)

• How it works:
  • Routers share full routing tables with neighbors periodically (every 30 sec in RIP).
  • Uses hop count as the metric (max 15 hops).
• Example: RIP in a Small Office

• Problems:
  • Slow convergence (takes time to update routes).
  • Routing loops (e.g., "count to infinity" where routers keep advertising a bad route).
• Solutions:
  • Split Horizon: Don’t advertise a route back to where it came from.
  • Poison Reverse: Advertise a bad route with a cost of 16 (infinity).
  • Hold-Down Timers: Suppress updates for a route that just went down.

2. Link-State Routing (OSPF)

• How it works:
  • Each router floods its entire topology (links and costs) to all routers.
  • Uses Dijkstra’s algorithm to compute the shortest path.
• Example: OSPF in NTC’s Backbone

• Advantages:
  • Fast convergence (updates propagate quickly).
  • No loops (each router has a complete topology map).
• Disadvantages:
  • High CPU/memory usage (storing full topology).
  • Complex to configure.

Comparison Table: RIP vs. OSPF

---

Subnetting: Dividing Networks for Efficiency

Subnetting splits a larger network into smaller subnets to:

• Reduce broadcast traffic.
• Optimize IP address allocation.
• Improve security (isolate departments).

How Subnetting Works:

1. Borrow bits from the host portion of the IP to create subnets.
2. Subnet Mask defines the new network/host boundary.
   • Example: 255.255.255.128 = /25 (borrowed 1 bit).

Worked Example: Subnetting 192.168.20.0/24

Given:

• Network: 192.168.20.0/24 (Class C)
• Subnet mask: 255.255.255.128 (/25)

Steps:

1. Determine borrowed bits:
   • Original mask: 255.255.255.0 (/24)
   • New mask: 255.255.255.128 (/25) → 1 borrowed bit.
2. Calculate subnets:
   • 2^1 = 2 subnets.
   • Subnet 1: 192.168.20.0/25 (range: .0 to .127)
   • Subnet 2: 192.168.20.128/25 (range: .128 to .255)
3. Hosts per subnet:
   • 2^(32-25) - 2 = 126 hosts per subnet (subtract 2 for network/broadcast).

Visual:

+-------------------+-------------------+
| Subnet 1          | Subnet 2          |
| 192.168.20.0/25   | 192.168.20.128/25 |
| Hosts: .1 to .126  | Hosts: .129 to .254|
+-------------------+-------------------+

Real-World Example: eSewa’s Payment Routing

• Problem: eSewa’s servers in Kathmandu and Pokhara need to communicate with banks (e.g., NMB, Global IME) without flooding the entire network with broadcast traffic.
• Solution: Subnetting divides the network:
  • 10.0.0.0/24 → Subnet for Kathmandu servers (10.0.0.0/25).
  • 10.0.0.128/25 → Subnet for Pokhara servers (10.0.0.129/25).
  • 10.0.1.0/24 → Subnet for bank connections.

---

Routing Loops: The Silent Network Killer

A routing loop occurs when:

• Packet A → Router 1 → Router 2 → Router 1 → Router 2 → ... (infinite loop).
• Cause: Misconfigured static routes or dynamic routing protocols without safeguards.

How to Prevent Routing Loops:

1. Split Horizon: Don’t advertise a route back to the neighbor it was learned from.
   • Example: If Router A learns 10.0.0.0/24 from Router B, it won’t advertise 10.0.0.0/24 back to Router B.
2. Poison Reverse: Advertise a bad route with a high cost (16).
   • Example: If Router A detects a loop to 10.0.0.0/24, it advertises it with cost 16 to neighbors.
3. Hold-Down Timers: Suppress updates for a route that just went down for a set time (e.g., 60 sec).

Example: RIP Routing Loop

---

In the Real World

1. Ncell’s Core Network

   • Uses OSPF for dynamic routing between cities (Kathmandu, Pokhara, Biratnagar).
   • TTL fields ensure packets don’t loop if a path fails.
   • Example: When you call a number in Pokhara, your packet’s TTL starts at 64 and decrements at each hop (Kathmandu router → Pokhara router).

2. eSewa’s Payment Gateway

   • Subnetting isolates payment processing servers from customer-facing servers.
   • Example: The 10.0.0.0/24 network is split into:
     • 10.0.0.0/25 (Customer web servers)
     • 10.0.0.128/25 (Payment processing servers)
     • 10.0.1.0/24 (Bank connections)
   • Static routes ensure payments go directly to banks without unnecessary hops.

3. Daraz’s Order Fulfillment

   • Dynamic routing (BGP) connects Daraz’s warehouses to NTC’s fiber backbone.
   • Example: When you order from Daraz, your request:
     1. Goes to Daraz’s server (IP: 203.123.45.100).
     2. Uses BGP to find the fastest path to the nearest warehouse.
     3. The packet’s TTL ensures it doesn’t loop if a path fails.

---

Exam Tip

1. IPv4 Header Questions:

   • Always draw the header and label all 13 fields. Examiners love this!
   • TTL and checksum are high-yield topics. Know how they prevent loops and errors.
   • Example Question: "Explain the role of the Protocol field in the IPv4 header." → Answer: It tells the destination which upper-layer protocol (TCP/UDP/ICMP) to pass the data to.

2. Routing Algorithms:

   • Compare RIP and OSPF in a table (metric, updates, loops, scalability).
   • Draw a sequence diagram for RIP updates or a state diagram for OSPF flooding.
   • Example Question: "Why does RIP suffer from slow convergence?" → Answer: It uses periodic updates (every 30 sec) and doesn’t detect topology changes immediately.

3. Subnetting Calculations:

   • Memorize the formula:
     • Number of subnets = 2^borrowed_bits.
     • Hosts per subnet = 2^(32 - subnet_mask_bits) - 2.
   • Practice with real IPs:
     • Given 192.168.1.0/26, calculate subnets and hosts.
     • Answer: 4 subnets, 62 hosts each.
   • Example Question: "Calculate subnets and hosts for 172.16.0.0/20." → Show your work step-by-step.

4. Real-World Applications:

   • Link to Nepalese companies:
     • Ncell: OSPF for core routing.
     • eSewa: Subnetting for security.
     • Daraz: BGP for global connectivity.
   • Example Question: "How does Ncell prevent routing loops?" → Answer: Uses OSPF (link-state) and split horizon in legacy RIP networks.

5. Common Pitfalls:

   • Forgetting to subtract 2 for network/broadcast in host calculations.
   • Misaligning subnet masks (e.g., 255.255.255.240 is /28, not /24).
   • Not recalculating checksums after modifying TTL (examiners test this!).

---

Practice Questions (Based on Past Exams)

1. IPv4 Header:

   • "Explain the IPv4 header format in detail." → Draw the header and describe each field’s role (e.g., TTL prevents loops, checksum ensures integrity).

2. Routing:

   • "Differentiate between static and dynamic routing with a suitable example."
     • Static: Manually configured (e.g., home router to ISP).
     • Dynamic: Routers exchange info (e.g., Ncell’s OSPF between cities).

3. Subnetting:

   • "Calculate total number of subnets and hosts per subnet for Network Address 192.168.20.0 and subnet mask 255.255.255.128."
     • Answer: 2 subnets, 126 hosts each.

4. Routing Algorithms:

   • "Explain Distance Vector Routing with an example and compare it with Link State Routing."
     • Use the RIP sequence diagram and OSPF state diagram above.

5. Real-World Scenario:

   • "How would you subnet the network 172.16.0.0/16 to support 5 departments, each needing 200 hosts?"
     • Answer: Borrow 3 bits (/19), creating 8 subnets (5 used), 8,190 hosts each (but only 200 used per subnet). Use 172.16.0.0/19 to 172.16.32.0/19.