IP Subnet Calculator

The number after the slash counts the bits at the front of the address that identify the network; whatever is left identifies hosts inside it. A /24 fixes twenty-four bits and leaves eight, so the block holds 256 addresses and the final octet is yours to hand out. Each step up halves the block — /25 is 128 addresses, /26 is 64, /27 is 32 — and each step down doubles it. The prefix carries the same information as a dotted-decimal mask, so /24 and 255.255.255.0 say the same thing; most equipment accepts either, but mixing the two forms in your head is where slips begin.

Because the first address in a block names the network itself and the last is the broadcast address, which reaches every host on the segment at once, so neither can sit on an interface. Plan for a third loss as well: the router or default gateway takes an address out of the same block, conventionally the first or last usable one. A /29 of eight addresses therefore supports five machines and a gateway, not eight. That overhead barely registers on a /24 and dominates a /29, which is why small blocks feel so much smaller than their size suggests.

Counting down from a /24 and its 254: a /25 gives 126, /26 gives 62, /27 gives 30, /28 gives 14, /29 gives 6 and /30 gives 2. Going the other way, /23 gives 510 and /22 gives 1,022. Pick with growth in mind, because resizing later means renumbering every static address, every firewall rule and every DHCP reservation that names the old range. A /24 is the common default for an office floor not because it is correct but because it is roomy and easy to read at a glance.

A wildcard mask is the subnet mask inverted: where the mask has ones, the wildcard has zeros. A /26 has the mask 255.255.255.192 and the wildcard 0.0.0.63. Read it as a statement of which bits may vary — zero means the bit must match, one means ignore it. Cisco access lists and OSPF network statements are written this way, and entering a subnet mask where a wildcard is expected is a classic error that either matches nothing or matches far more than you meant. Both forms appear in the results, so you can take whichever the box in front of you is asking for.

Three ranges are set aside by RFC 1918 for private use and are not routed across the public internet: 10.0.0.0 to 10.255.255.255, 172.16.0.0 to 172.31.255.255, and 192.168.0.0 to 192.168.255.255. Others are reserved for particular jobs — 127 for loopback, 169.254 for the link-local address a machine assigns itself when DHCP fails. Private ranges can be used by every organisation at once, which is why they are translated at the border and why two companies merging so often find that both built on 10.0.0.0/8. The scope shown with the results says which category an address falls into.

The usual cause is a mask mismatch rather than a cable. A machine decides whether a destination is local by masking its own address and the destination's with its own subnet mask and comparing the two results; when they differ, the packet goes to the default gateway instead of onto the wire. Give one host a /24 and its neighbour a /25 on the same segment and you get an asymmetry — one believes the other is local, the other does not — which appears as traffic that works in one direction only. Check that every host carries the same prefix and gateway, and that the gateway sits inside the block.

Nothing leaves the tab. The calculation is arithmetic on a single 32-bit number performed by the page's own JavaScript, and the component makes no request of any kind — there is no server route behind it, so there is nothing that could log the ranges you type. Nothing is stored either: what you enter lives in ordinary component state, and closing the tab is the whole of the cleanup. Being deliberate about this is reasonable in general. An addressing plan describes where things sit on your network, and pasting one into a site that does its sums on a server hands over a small piece of a map you would not otherwise publish.

By halving. A /24 splits into two /25s at .0 and .128, four /26s at .0, .64, .128 and .192, eight /27s in steps of thirty-two, and onwards. Each child block starts at a multiple of its own size, which is the rule that makes the boundaries predictable once you know the step. Where departments need different sizes, allocate the largest block first and carve the remainder into smaller ones, so nothing has to start on a boundary it cannot occupy. You work one block at a time here: enter a candidate, read its range, then enter the next to see where it begins. No list is generated for you.