VLSM Calculator

Variable Length Subnet Masking divides one IPv4 block into subnets of different sizes, choosing each mask from the number of hosts that subnet actually needs. For a requirement of H hosts, the allocation needs H plus 2 addresses, because the all-zeros address in the block is the network identifier and the all-ones address is the directed broadcast. The smallest power of two that covers that total gives the host-bit count n, the prefix is 32 minus n, and the usable capacity is 2 to the power of n minus 2. A department of 60 hosts therefore needs 62 addresses, which rounds up to 64, a /26 with mask 255.255.255.192 and 62 usable addresses. Allocation order matters because every subnet must begin on an address that is an exact multiple of its own size: a /26 can start only at .0, .64, .128 or .192, while a /29 may start at any multiple of 8. Placing the largest blocks first keeps the remaining space aligned for the smaller ones; taking them in arbitrary order can strand usable addresses even when the arithmetic total still fits. A /30 gives only 2 usable addresses and is the classic point-to-point link size, while RFC 3021 allows /31 links with no broadcast address on routers that support it. A /24 caps out at 254 hosts, and pushing much beyond that in one broadcast domain raises ARP load rather than saving addresses. VLSM also assumes the routing protocol carries the mask with each prefix, so summarization boundaries should be planned before the addresses are handed out.

IPv4 subnetting was defined by Jeffrey Mogul and Jon Postel in RFC 950 in August 1985, which let a single classful network number be split internally using a subnet mask. That first scheme assumed one mask for the whole network. RFC 1009, the 1987 requirements document for Internet gateways, explicitly described subnets of differing sizes within one network, and RFC 1878 in 1995 published the reference table of prefix lengths, masks, and host counts that administrators used for years. The idea only became practical, however, once routing protocols could advertise a mask alongside each prefix. RIP version 1 and IGRP could not, so early networks were locked into a single mask; RIP version 2 in RFC 1723 in 1994, OSPF, and Cisco's EIGRP carried the mask and made variable-length designs deployable. The pressure behind all of this was address exhaustion and routing-table growth in the early 1990s. The IETF ROAD group work led to Classless Inter-Domain Routing, specified in RFC 1518 and RFC 1519 in September 1993, which discarded the Class A, B, and C boundaries entirely and made the prefix length an explicit part of every route. VLSM is the same classless arithmetic applied downward inside an organization, while CIDR aggregation applies it upward between providers. RFC 1812 in 1995 then made classless forwarding a requirement for IPv4 routers, and the largest-first allocation method taught today comes from this period of network design practice.