VLSM Calculator

How to use this calculator

1. Enter the base networkIn CIDR notation — address and prefix. The calculator normalises whatever you enter to the actual network address, so 192.168.1.37/24 becomes 192.168.1.0/24.
2. Enter usable hosts, not total addressesEnter the number of devices that need an address. The calculator adds two for the network and broadcast addresses itself, which is the step most often forgotten by hand.
3. Do not sort them yourselfEnter subnets in any order. The calculator sorts largest-first before allocating, because that ordering is what makes VLSM work at all.
4. Read the wasted columnIt shows how many usable addresses each block has beyond what you asked for. Large waste on a small subnet usually means the requirement sits just above a power of two.

Why the largest subnet has to go first

This is the rule that makes VLSM different from ordinary subnetting, and it is the one that trips people up in the CCNA exam.

Every subnet must begin on a boundary that is a multiple of its own block size. A /26 block holds 64 addresses, so it can only start at .0, .64, .128 or .192. It cannot start at .32, even if .32 happens to be free.

Allocate smallest-first and you break that rule immediately. Give a 10-host subnet the first 16 addresses, and the next free address is .16. A 128-address block cannot start at .16 — the next valid boundary is .128, so everything from .16 to .127 is stranded. You have wasted 112 addresses to save none.

Allocate largest-first and the boundaries fall into place automatically. A 128-block starts at .0 and ends at .127. The next free address is .128, which is a valid boundary for a 64-block. That ends at .191, and .192 is a valid boundary for a 32-block. Each block leaves the cursor on a boundary the next-smaller block can use.

That is the whole trick, and it is why the rule is stated as an ordering rather than as arithmetic. The arithmetic is trivial; the ordering is what makes it fit.

Worked example: 192.168.1.0/24 across four departments

Requirements of 100, 50, 25 and 10 hosts, already in descending order.

Sales needs 100 hosts. Add two for network and broadcast: 102. The next power of two is 128, so this is a /25 — 32 minus 7 host bits. It takes 192.168.1.0/25, usable range .1 to .126, broadcast .127. Usable addresses: 126, so 26 are wasted.

Engineering needs 50. Plus two is 52, next power of two is 64, a /26. It starts at .128, which is a multiple of 64. Range .129 to .190, broadcast .191. Usable 62, wasted 12.

Admin needs 25. Plus two is 27, next power of two is 32, a /27. Starts at .192, range .193 to .222, broadcast .223. Usable 30, wasted 5.

Servers needs 10. Plus two is 12, next power of two is 16, a /28. Starts at .224, range .225 to .238, broadcast .239. Usable 14, wasted 4.

Total allocated: 128 + 64 + 32 + 16 = 240 addresses. Sixteen remain free from 192.168.1.240, which is exactly one more /28 if a fifth department appears.

With classful subnetting into four equal blocks you would get four /26s of 64 addresses each. Sales, needing 100, would not fit at all. That is the case for VLSM in one sentence.

Block size reference

The /30 row is worth memorising separately. Two usable addresses is exactly what a point-to-point link between two routers needs, and using anything larger for one wastes address space at every hop. Some designs use /31 links, which dispense with the network and broadcast addresses entirely under RFC 3021, though not every platform supports them.

The mistakes that cost marks and cause outages

Forgetting the plus two is the most common error, and it hides until deployment. A requirement for exactly 30 hosts needs 32 addresses, which means a /27 gives you 30 usable — just enough. A requirement for 31 hosts needs 33, which pushes you to a /26 and 62 usable. The jump is not gradual, and a subnet sized from the host count alone will be one device short.

Sizing for today rather than for growth is the operational version of the same mistake. A subnet at 95% capacity cannot be expanded in place — the next block is already allocated to something else. Renumbering a live subnet is disruptive in a way that allocating a slightly larger block on day one is not. Sizing to roughly double the current requirement is a common convention.

Allocating in the order the requirements arrived, rather than by size, produces a design that looks fine on paper and fails to fit. Always sort first.

Overlapping ranges arise when blocks are placed by hand without checking boundaries. Two subnets whose ranges intersect will produce routing behaviour that is difficult to diagnose because it is intermittent — traffic reaches the wrong destination only for some addresses.

And summarisation is worth thinking about at design time rather than after. Allocating contiguous blocks to related subnets means they can be summarised into a single route advertisement later. Scattering them means they cannot, and the routing table grows for no benefit.

Assumptions and limitations

• IPv4 only — IPv6 subnetting follows different conventions
• Reserves two addresses per subnet for network and broadcast, so does not model RFC 3021 /31 point-to-point links
• Allocates contiguously from the start of the base network with no reserved gaps
• Does not check whether the base network overlaps anything already deployed
• Handles up to six subnets in one pass — larger designs are usually built hierarchically anyway
• Does not generate router configuration or check for summarisation opportunities
• Assumes classless routing throughout, which every modern platform supports

Questions people ask about this in Philippines