Lecture 8 - Network Layer (Part 2)

Content Overview

• IP Address
• Classless Inter-Domain Routing (CIDR)
• Variable Length Subnet Mask (VLSM)
• IPv6
• Network Address Translation (NAT)
• ICMP

IP Address: Network/Subnet

4 Main Techniques to Split IP Range

• IP address classful (A, B, C)
• Public/Private IP
• Classless Inter-Domain Routing (CIDR)
• Variable Length Subnet Mask (VLSM)

บอกแล้วว่าอย่าลืมนับระหว่าง Router นะ
ตัว Router จะเป็นตัวต่อระว่าง Network/Subnet
ถ้ามี Computer มาต่อกับ Network A 50 เครื่อง ก็ไม่ได้มี Subnet เพิ่มนะ

When to Use Which?

Design IPs for Network

Using Classful (Inefficient)

⚠️ Problem: Using classful wastes a massive amount of IPs. For example, Network B only needs 300 IPs but gets 65,534 — like renting an entire warehouse just to store a bicycle.

Subnetting a Class A/B/C Address

Class C (254 Hosts)

• Network Address: 192.168.3.0
• Usable IP range: 192.168.3.1 – 192.168.3.254 ($2^8 - 2$)
• Broadcast Address: 192.168.3.255
• Mask: 255.255.255.0 (/24)

Class B (65,534 Hosts)

• Network Address: 172.16.0.0
• Usable IP range: 172.16.0.1 – 172.16.255.254 ($2^{16} - 2$)
• Broadcast Address: 172.16.255.255
• Mask: 255.255.0.0 (/16)
• ⚠️ Too many unused IPs! → ทำไงอะ มีครึ่ง ๆ กลาง ๆ ระหว่าง /24 กับ /16 ได้ป้ะะะ??

Classless Inter-Domain Routing (CIDR)

CIDR is pronounced "cider" 🍎

What is CIDR?

• A method for allocating IP addresses and for IP routing
• Subnet portion of address has arbitrary length (not fixed to class boundaries)
• Address format: $a.b.c.d/x$ where $x$ = number of bits in subnet portion

$\boxed{\text{Address format: } a.b.c.d/x}$

Binary Example

11001000 00010111 00010000 00000000
|________ subnet part ________|host|
         200.23.16.0/23

Key Questions for Any Subnet

1. How many subnets does the chosen mask produce?
2. How many valid hosts per subnet?
3. What are the valid subnets (network addresses)?
4. What is the broadcast address of each subnet?
5. What are the valid hosts in each subnet?

#FinalExam → 192.168.100.0/24, we want to split into 5 subnet (How?)
What are the number of the host for each subnet

CIDR: Key Formulas

$\boxed{\text{Host bits} = 32 - \text{prefix length}}$
$\boxed{\text{Total usable IPs} = 2^{\text{host bits}} - 2}$
$\boxed{\text{Total subnets} = 2^{\text{subnet bits added}}}$

Think of it like slicing a pizza: the prefix length tells you how many slices you've already "locked in" as the network. The remaining bits are your slices to give to hosts. You always lose 2 slices — one for the network address and one for the broadcast.

CIDR Examples

บางครั้งคำถามจะถามว่า IP Address อันนี้เป็น Regular IP, Brodcast IP Address หรือว่าอะไร
เจอแน่ #FinalExam

• ตอน #FinalExam อย่าลืมเอาเข้าไปด้วย ช่วยนะ
  

Example 1: 128.143.137.144/20

• Host bits = 32 - 20 = 12
• Total IPs = $2^{12} - 2 = 4094$
• Network Address = 128.143.128.0/20
• First Host = 128.143.128.1
• Last Host = 128.143.143.254
• Broadcast = 128.143.143.255

Example 2: 192.168.10.0/25 (mask 255.255.255.128)

• Host bits = 32 - 25 = 7
• Total IPs per subnet = $2^7 - 2 = 126$
• Subnets:
  • 192.168.10.0/25 → Host: .1–.126, Broadcast: .127
  • 192.168.10.128/25 → Host: .129–.254, Broadcast: .255

CIDR Notation to Subnet Mask Reference

CIDR: Splitting a /24 into Smaller Subnets

Split 192.168.10.0/24 into 2 subnets → /25

• Subnets needed: 2 = $2^1$ → borrow 1 bit → mask becomes /25 (255.255.255.XXX)
• IPs per subnet: $2^7 = 128$ total, $128 - 2 = 126$ usable
• Valid subnets: 0, 128

Split 192.168.10.0/24 into 4 subnets → /26

• Subnets needed: 4 = $2^2$ → borrow 2 bits → mask becomes /26 (255.255.255.192)
• IPs per subnet: $2^6 = 64$ total, $64 - 2 = 62$ usable
• Valid subnets: 0, 64, 128, 192

Split 192.168.10.0/24 into 5 subnets → /27

• Subnets needed: 5 → round up to $2^3 = 8$ → borrow 3 bits → mask /27 (255.255.255.224)
• IPs per subnet: $2^5 = 32$ total, $32 - 2 = 30$ usable
• Valid subnets: 0, 32, 64, 96, 128, 160, 192, 224

Design IPs for Network (CIDR)

Given 192.168.0.0/20, use CIDR → /23

• Max hosts = 300 → $2^9 = 512$ IPs → /23
• Total subnets = $2^3 = 8$

⚠️ Still wasteful — every subnet gets 510 IPs regardless of actual need. This is where VLSM shines.

เวลาใช้ CIDR ต้องใช้ /XX เท่ากัน ถูกป้ะ ตรงระหว่าง Router เปลืองมาก ใช้แค่ 2 IPs ทำยังไงให้มัน Efficient กว่านี้ได้มั้ย

โจทย์ให้มาแบบนี้ ไม่ให้ CIDR Prefix มาทำไง

• เช็คก่อนว่ามีกี่ Subnet → 5 ใช่ม้า
• เช็ค maximum number of host required → 300
• เอายังไงให้พอ ก็เลือกเองได้เลย รู้เองได้เลย /23, /22 (ก็ได้นะ แต่เปลืองว่ะ)

Variable Length Subnet Mask (VLSM)

What is VLSM?

• An extension of CIDR that allows different subnets to use different prefix lengths
• Each subnet gets exactly (or closely) what it needs
• Key rule: Assign the largest subnet first

VLSM is like cutting a rope into pieces of different sizes for different purposes, instead of cutting equal chunks and wasting the shorter ones.

VLSM vs CIDR Comparison

Given network: 204.15.5.0/24 with 5 subnets (netA=14, netB=28, netC=2, netD=7, netE=28)

With CIDR (/27 for all):

• 5 subnets used, each with 30 hosts
• 3 unused subnets available
• Unused IPs: 161 | Available: 3 subnets = 90 hosts

With VLSM (size-matched):

• Unused IPs: 145 | Available: 11 subnets = 134 hosts
• Far more efficient!

CIDR /XX ต้องเท่ากันหมด
แต่ VLSM it can be different based on the number of IP required!

VLSM Step-by-Step Example

Network: 204.15.5.0/24

Requirements:

• netA: 14 hosts
• netB: 28 hosts
• netC: 2 hosts
• netD: 7 hosts
• netE: 28 hosts

Step 1: Determine masks needed per subnet

Step 2: Assign largest first

204.15.5.0/24 → /27

• Hosts = $2^5 - 2 = 30$
• Subnets = $2^3 = 8$: {0, 32, 64, 96, 128, 160, 192, 224}

204.15.5.64/27 → /28 (split for netA, netD)

• Hosts = $2^4 - 2 = 14$
• Subnets = $2^{(28-27)} = 2$: {64, 80}

204.15.5.96/27 → /30 (split for netC)

• Hosts = $2^2 - 2 = 2$
• Subnets = $2^{(30-27)} = 8$: {96, 100, 104, 108, 112, 116, 120, 124}

Step 3: Final Assignment

Design IPs for Network (VLSM)

Given 192.168.0.0/20, use VLSM

Requirements:

• Network A: 100 hosts
• Network B: 300 hosts
• Network C: 254 hosts
• R1-R2 link: 2 hosts
• R2-R3 link: 2 hosts

Assign largest first:

Step breakdown:

• 192.168.0.0/20 → /23: Hosts = $2^9 - 2 = 510$, Subnets = $2^3 = 8$ → [0,2,4,6,8,10,12,14]
• 192.168.2.0/23 → /24: Hosts = $2^8 - 2 = 254$, Subnets = $2^{(24-23)} = 2$ → [2,3]
• 192.168.3.0/24 → /25: Hosts = $2^7 - 2 = 126$, Subnets = $2^{(25-24)} = 2$ → [0,128]
• 192.168.3.128/25 → /30: Hosts = $2^2 - 2 = 2$, Subnets = $2^{(30-25)} = 32$ → [128,132,136,…,252]

IP Addressing: How to Get One?

• Q: How does a network get its subnet part?
• A: Gets allocated a portion of its provider ISP's address space

Example: ISP Block Allocation

• ISP block: 200.23.16.0/20
• ISP splits into 8 /23 blocks for organizations:

Hierarchical Addressing: Route Aggregation

• ISP advertises a single prefix (200.23.16.0/20) to the internet
• This covers all 8 organizations efficiently
• Hierarchical addressing allows efficient advertisement of routing information

ISP แต่ละเจ้าก็จะมี AS Number, เป็น Unique Identifier ของแต่ละเจ้า

IPv4 Address Exhaustion

• ICANN allocated the last chunk of IPv4 addresses in 2011
• IPv4 = 32-bit = ~4.3 billion addresses (not enough!)
• Solutions:
  • NAT (short-term workaround) → separate public range, and private range
    • ถ้า network จะ go outside ก็ต้องใช้ NAT แปลงเป็น Public IP
  • IPv6 (long-term solution)

IPv6

Motivation

มันไม่มี Fixed date ไง ว่าจะเปลี่ยนไปใช้วันนี้ ๆ, hardware บางอย่างก็ไม่รองรับ
เราก็เลยต้องใช้ IPv4 and IPv6 ร่วมกันไป

• IPv4 32-bit address space nearly exhausted
• IPv6 has 128-bit address space
• Fixed-length 40-byte header for faster processing/forwarding
• Enables different network-layer treatment of "flows"

IPv6 Datagram Format

|-------- 32 bits --------|
| ver | pri |  flow label  |
| payload len | next | hop |
|    source address       |
|      (128 bits)         |
|   destination address   |
|      (128 bits)         |
|     payload (data)      |

• ver: IP version (6)
• pri (priority): identify priority among datagrams in same flow
• flow label: identify datagrams in same "flow" (not well defined)
• next hdr: identifies upper-layer protocol
• hop limit: replaces TTL

What's Missing Compared to IPv4?

• ❌ No checksum (speeds up router processing)
• ❌ No fragmentation/reassembly (done at endpoints only)
• ❌ No options (available as next-header extension)

IPv6 is like a streamlined express lane — it removed all the extra checks IPv4 had at every router to make packet forwarding blazing fast.

IPv4 to IPv6 Transition

อาจจะถามใน #FinalExam นะ พาร์ทข้อเขียน

Challenge

• Not all routers can be upgraded simultaneously
• No "flag day" where everything switches at once
• Mixed IPv4 and IPv6 routers must coexist

Solution: Tunneling

• IPv6 datagram carried as payload inside an IPv4 datagram between IPv4 routers
• Called "packet within a packet"
• Also used in 4G/5G networks

IPv4 datagram:
+---IPv4 header (src:B, dst:E)---+
|  IPv6 datagram:                |
|  +-- IPv6 header (src:A,dst:F) |
|  |   payload (data)            |
|  +-----------------------------+
+--------------------------------+

Tunneling Flow (A→B→C→D→E→F)

• A→B: Native IPv6
• B→C→D→E: IPv6 inside IPv4 tunnel (B wraps, E unwraps)
• E→F: Native IPv6

IPv6 tunneling through IPv4 is like putting an international package into a domestic shipping box — the outer box travels through the local system, but inside is the real package destined internationally.

NAT: Network Address Translation

Translate private to public IP addresses.

What is NAT?

• All devices in a local network share just one public IPv4 address as seen by the outside world
• Devices inside use private IP addresses

• WAN side (Public), LAN side (Private)
• ใน NAT trasnaltion table ไม่ได้มีแค่ (Mapping) IP อย่างเดียว
  • ไม่งั้น Public IP outside จะมีแค่อันเดียวอะดิ
  • เราต้องมี Port ด้วย → It allows us to map in the table around (16 bits = 65k)

Private IP Address Ranges

NAT is like an apartment building: many residents (private IPs) share one street address (public IP). The lobby receptionist (NAT router) knows which apartment each incoming parcel goes to via a mapping table.

NAT Advantages

• Only one public IP needed from ISP for all internal devices
• Can change internal addresses without notifying outside world
• Can change ISP without reconfiguring internal devices
• Security: internal devices not directly addressable from outside

How NAT Works

Outgoing Datagram (LAN → WAN): (Internal → External (Public Internet))

1. Replace (source IP, source port) → (NAT IP, new port)
   • Replace at the packet!
2. Record mapping in NAT translation table

เก็บข้อมูลเหล่านี้ทำไม? ใน NAT Translation Table: ก็เวลามีของเข้ามา มันก็จะเช็คในตารางก่อนไง แล้วถึงจะ forward ไปถูก

Incoming (WAN → LAN):

1. Look up (NAT IP, port) in translation table
2. Replace with stored (original IP, original port)
3. Forward to correct internal host

NAT Example

Host 10.0.0.1:3345 → 128.119.40.186:80

Step 1: Host sends datagram
  S: 10.0.0.1:3345  D: 128.119.40.186:80

Step 2: NAT router rewrites source
  S: 138.76.29.7:5001  D: 128.119.40.186:80
  [NAT table: 138.76.29.7:5001 ↔ 10.0.0.1:3345]

Step 3: Reply arrives
  S: 128.119.40.186:80  D: 138.76.29.7:5001

Step 4: NAT router rewrites destination
  S: 128.119.40.186:80  D: 10.0.0.1:3345

NAT Controversies

• ⚠️ Routers "should" only process up to Layer 3 (NAT touches Layer 4 port numbers)
• ⚠️ Address shortage "should" be solved by IPv6
• ⚠️ Violates end-to-end argument (port manipulation by network device)
• ⚠️ NAT traversal problem: hard for external client to connect to server behind NAT
• ✅ But NAT is widely used in home, institutional, and cellular networks

ICMP: Internet Control Message Protocol

What is ICMP?

• Used by hosts and routers to communicate network-level information
• Sits logically above IP (carried inside IP datagrams)
• Handles:
  • Error reporting (unreachable host/network/port/protocol)
  • Diagnostic tools (ping, traceroute)

ICMP Message Format

• Type + Code + first 8 bytes of the offending IP datagram

Common ICMP Messages

Traceroute and ICMP

How Traceroute Works

1. Source sends sets of UDP segments to destination
   • 1st set: TTL = 1
   • 2nd set: TTL = 2
   • $n$th set: TTL = $n$
2. When TTL expires at nth router:
   • Router discards the datagram
   • Sends back ICMP TTL Expired message (Type 11, Code 0)
   • Message may include router name and IP
3. Source records RTT for each hop

Stopping Criteria

• ✅ UDP segment reaches destination host
• ✅ Destination returns ICMP Port Unreachable (Type 3, Code 3)
• ✅ Source stops

Traceroute is like shouting "Marco Polo" with a timer that resets at each pool wall. Each router along the way shouts back "Polo!" when the timer runs out, telling you how far away it is.

TTL = Time To Live

Summary