Subnetting

“Subnetting is the art of dividing a network into smaller, manageable pieces.”

Overview

Subnetting divides a larger IP network into smaller sub-networks (subnets). It improves network performance (smaller broadcast domains), enhances security (isolate departments), and enables efficient address allocation.

Why Subnet?

graph TD
    subgraph "Before Subnetting"
        N["Network 192.168.1.0/24<br/>254 hosts, one broadcast domain"]
        H1[Host A] --- N
        H2[Host B] --- N
        H3[Host C] --- N
        H4[Host D] --- N
        Note1["All hosts see all broadcasts<br/>Security risk<br/>Performance issues"]
    end
    
    subgraph "After Subnetting"
        S1["192.168.1.0/26<br/>Engineering<br/>62 hosts"]
        S2["192.168.1.64/26<br/>Marketing<br/>62 hosts"]
        S3["192.168.1.128/26<br/>Finance<br/>62 hosts"]
        S4["192.168.1.192/26<br/>HR<br/>62 hosts"]
        Note2["Isolated broadcast domains<br/>Better security<br/>Improved performance"]
    end

How Subnetting Works

Step-by-Step Process

Example: Divide 192.168.1.0/24 into 4 equal subnets

1. Determine bits needed: 4 subnets → 2 bits (2² = 4)
2. New prefix length: /24 + 2 = /26
3. New subnet mask: 255.255.255.192
4. Host bits: 32 - 26 = 6 → 2^6 - 2 = 62 hosts per subnet

Subnet Calculation

Original: 192.168.1.0/24
Binary:   11000000.10101000.00000001.00000000
Mask:     11111111.11111111.11111111.00000000 (/24)

New Mask: 11111111.11111111.11111111.11000000 (/26)
Subnet bits: ^^ (2 bits from host portion)

Resulting Subnets

Subnetting Formulas

Number of subnets = 2^n (n = bits borrowed from host portion)
Hosts per subnet = 2^h - 2 (h = remaining host bits)
Subnet mask = original mask + n bits

Network address = IP AND Mask
Broadcast address = Network OR (NOT Mask)
First usable = Network + 1
Last usable = Broadcast - 1

Variable Length Subnet Masking (VLSM)

VLSM allows different subnet sizes within the same network — no wasted addresses.

Example: 192.168.1.0/24 divided for different needs:

graph TD
    P["192.168.1.0/24<br/>254 hosts total"] --> E["192.168.1.0/25<br/>Engineering: 126 hosts"]
    P --> M["192.168.1.128/26<br/>Marketing: 62 hosts"]
    P --> F["192.168.1.192/27<br/>Finance: 30 hosts"]
    F --> W1["192.168.1.224/30<br/>WAN Link 1: 2 hosts"]
    F --> W2["192.168.1.228/30<br/>WAN Link 2: 2 hosts"]

Practical Subnetting Examples

Example 1: How many subnets and hosts?

Given: 172.16.0.0/20

Subnets: The /20 means 20 bits for network (vs default /16 for Class B)
Subnet bits borrowed: 20 - 16 = 4
Number of subnets: 2^4 = 16
Host bits: 32 - 20 = 12
Hosts per subnet: 2^12 - 2 = 4,094

Example 2: What subnet is this host on?

Given: Host 10.45.67.129/18

Mask: 255.255.192.0 (/18)
Binary mask: 11111111.11111111.11000000.00000000

IP:  00001010.00101101.01000011.10000001
Mask:11111111.11111111.11000000.00000000
AND: 00001010.00101101.01000000.00000000

Network: 10.45.64.0/18
Broadcast: 10.45.127.255
Host range: 10.45.64.1 - 10.45.127.254

Example 3: Design a network

Requirement: 5 offices, each needs 50 hosts

50 hosts → need at least 50 + 2 = 52 addresses → /26 (62 hosts) works
Subnet mask: 255.255.255.192

Allocate from 10.0.0.0:
Office 1: 10.0.0.0/26    (hosts: .1 - .62)
Office 2: 10.0.0.64/26   (hosts: .65 - .126)
Office 3: 10.0.0.128/26  (hosts: .129 - .190)
Office 4: 10.0.0.192/26  (hosts: .193 - .254)
Office 5: 10.0.1.0/26    (hosts: .1 - .62)

Quick Subnetting Tricks

Powers of 2 Table

Magic Number Method

For quick subnetting, use the “magic number” (the value of the last network bit):

/26 → Last network bit is at position 64 (2^6)
Subnets: 0, 64, 128, 192
Each subnet has 64 addresses (62 usable)

/27 → Magic number = 32 (2^5)
Subnets: 0, 32, 64, 96, 128, 160, 192, 224
Each subnet has 32 addresses (30 usable)

Interview Questions

Beginner

Q1: What is subnetting and why is it needed? Subnetting divides a large network into smaller sub-networks. It’s needed for: (1) reducing broadcast traffic (smaller broadcast domains), (2) improving security (isolate departments), (3) efficient address allocation (don’t waste addresses), (4) easier network management.

Q2: How do you calculate the number of hosts in a subnet? Formula: 2^n - 2, where n = number of host bits (32 - prefix length). For example, /24 has 32-24=8 host bits → 2^8 - 2 = 254 hosts. We subtract 2 because the first address (all host bits 0) is the network ID and the last (all host bits 1) is the broadcast address.

Q3: What is the difference between subnet mask and CIDR notation? They represent the same thing in different formats. Subnet mask uses dotted decimal (255.255.255.0), CIDR uses the number of network bits (/24). Both mean the first 24 bits are the network portion.

Intermediate

Q4: Explain VLSM and its advantage over FLSM.

• FLSM (Fixed-Length): All subnets use the same mask. If the largest subnet needs 100 hosts, all subnets get /25 (126 hosts), even if most only need 10. Wastes addresses.
• VLSM (Variable-Length): Each subnet can have a different mask based on need. A subnet needing 100 hosts gets /25, one needing 10 gets /28. No waste.

Q5: How would you subnet 10.0.0.0/8 for a company with 200 branches, each needing 100 hosts?

1. 100 hosts → need 2^7 - 2 = 126 → /25 per branch
2. 200 branches × 128 addresses = 25,600 addresses
3. Allocate: 10.{branch}.0.0/25, 10.{branch}.0.128/25, 10.{branch}.1.0/25, etc.
4. Or use 10.{branch}.{subnet}.{host} with /25 subnets
5. Leaves plenty of room for growth within 10.0.0.0/8

Q6: Given 172.16.5.0/24, create subnets for departments needing 50, 30, 10, and 5 hosts.

Advanced / FAANG-Level

Q7: Design IP addressing for a data center with 500 servers across 10 racks. Design:

1. Supernet: 10.10.0.0/16 (65,534 addresses)
2. Per rack: 10.10.{rack}.0/24 (254 hosts each)
3. Management VLAN: 10.10.200.0/24 (out-of-band management)
4. Storage VLAN: 10.10.201.0/24 (iSCSI/NFS)
5. vMotion VLAN: 10.10.202.0/24 (live migration)
6. WAN/Internet: Public IPs on DMZ, NAT for internal
7. Loopbacks: 10.10.255.0/24 (router/switch loopbacks)
8. Summary route: Advertise 10.10.0.0/16 to WAN

Q8: How would you troubleshoot a host that can’t reach a host on a different subnet? Systematic approach:

1. Verify local config: Check IP, mask, default gateway (ip addr, ip route)
2. Test local connectivity: Ping own gateway
3. Check ARP: Is the gateway’s MAC resolved? (arp -a)
4. Test remote subnet: Ping remote host (may fail if routing issue)
5. Trace route: traceroute to see where packets stop
6. Check routing: Is there a route to the destination? (ip route get dest)
7. Check firewall: iptables -L or nftables list ruleset
8. Check VLAN: Are both hosts on correct VLANs? (switch config)
9. Check ACLs: Router/switch ACLs blocking traffic?

Q9: Explain supernetting and route aggregation. Supernetting (route aggregation) combines multiple smaller routes into one larger route:

192.168.0.0/24
192.168.1.0/24
192.168.2.0/24
192.168.3.0/24
→ Aggregated: 192.168.0.0/22 (one route instead of four)

Benefits:

• Smaller routing tables: Fewer entries = faster lookups
• Reduced BGP table size: Internet routing table has 900k+ prefixes; aggregation helps
• Simplified management: One route covers many subnets

Rules for aggregation:

• Networks must be contiguous
• Must align on power-of-2 boundaries
• All networks must share the same prefix

Common Mistakes

1. ❌ Forgetting to subtract 2 for usable hosts (network + broadcast addresses)
2. ❌ Not aligning subnets on power-of-2 boundaries
3. ❌ Using /31 for non-point-to-point links (only 0 usable hosts)
4. ❌ Confusing “number of subnets” with “number of hosts”
5. ❌ Not considering future growth when sizing subnets

Summary

• Subnetting divides large networks into smaller, manageable subnets
• Formula: Subnets = 2^n, Hosts = 2^h - 2
• VLSM allows different subnet sizes for efficient allocation
• Supernetting combines routes for smaller routing tables
• Subnet mask defines the boundary between network and host portions
• Always plan for growth when designing subnets

Cross-References

• IPv4 — IPv4 addressing fundamentals
• CIDR — Classless addressing notation
• NAT — Connecting private subnets to Internet
• DHCP — Automatic IP assignment within subnets

Cross References

• IPv4
• CIDR
• IP Protocol
• DHCP