Variable Length Subnet Masking, or VLSM, allow network administrators to create customized, less wasteful networks while keeping future network expansion in mind, according to the TCP/IP Guide. This is especially important in a business situation, where a varying number of IP addresses are needed for each department or location.
With traditional subnet calculations, each resulting network contains the same number of usable host IP addresses. While the needs of the largest LAN in the topology are met, blocks of unused addresses go to waste in smaller areas. This can be seen as leaving room for future growth but the IP addresses are limited to the existing subnet. If a new department or branch opens up, the business cannot just move the unused IP addresses to a new network. This results in fewer subnets available for future use, limiting the growth of the network. VLSM tries to prevent this by getting as close to the needed number of IP addresses for each LAN as possible, with few or no addresses left over, according to CCNA ICND2 Cert Guide: Variable Length Subnet Masks by Wendell Odom.
In a VLSM scheme, the network is first subnetted to meet the host needs of the largest LAN, then the subnets are further divided to accommodate smaller networks with as little waste of IP addresses as possible, according to Tutorials Point. Each new subnetwork is created using a different subnet mask, hence the term "variable length." This allows network administrators to create anything from large LANS requiring hundreds of host addresses to small links between intermediary devices where only two addresses are needed.