Subnetting

Every device (host) that talks on a network needs an IP address. A subnet (sub-network) is a logical division of a network based on IP addressing — a "network within a network." Grouping hosts into subnets (e.g., a Marketing subnet vs. a Sales subnet) keeps traffic organized, isolates broadcast domains, and improves efficiency.

Subnetting works by borrowing bits from the host portion of an IP address and reassigning them to identify a subnet. This splits an IP address into three logical pieces instead of two:

Before CIDR, IP address space was divided into "classes," each with a fixed default subnet mask. Classes aren't used to route traffic anymore, but understanding them makes the borrowing math easier to picture.

A subnet mask is a 32-bit number that marks which bits of an IP address are the network/subnet portion versus the host portion. It "masks out" the borrowed host bits. If you remember nothing else about subnetting, remember this — it's the foundation of the whole topic.

A mask/address pair can be written two ways:

The /25 is the CIDR prefix — it's shorthand for "25 bits of this 32-bit address are network/subnet bits." Reading a CIDR example:

The last-octet values below are the ones you'll use constantly — they apply whenever you're subnetting inside a single octet (the most common case on the exam and in the worked example below). Subnet count assumes borrowing starts from a /24 base.

IP classes aren't used to route traffic anymore — CIDR replaced them. CIDR applies the same bit-borrowing logic in reverse: instead of breaking one network into many subnets, it can aggregate (combine) multiple smaller networks into one larger block. This is called supernetting.

CIDR's real payoff is efficient address allocation — a network admin can hand out a block sized much closer to what's actually needed instead of being stuck with fixed classful sizes.

When a network is assigned more than one subnet mask, it's said to use a Variable Length Subnet Mask (VLSM) — this is what happens when you "subnet a subnet." VLSM lets each subnet be sized for exactly how many hosts it needs, instead of forcing every subnet to use the same fixed mask, which reduces wasted addresses.

Build a quick "cheat sheet" before subnetting any /24 network. Write the powers of two across the top, then stack a Subnets row, a Hosts row, and a CIDR row underneath:

The process, step by step:

192.168.1.0/25 splits one Class C network into two subnets of 126 hosts each (1 bit borrowed → 2 subnets).

Requirement: 4 subnets → borrow 2 bits (2² = 4) → new mask is /26 (255.255.255.192). Remaining 6 host bits → 2⁶ = 64 total addresses per subnet, block size 64, usable hosts = 64 − 2 = 62 (plenty of headroom over the 3 needed).