Who exactly determines the most efficient way to divide an IP address block when different departments have vastly different connectivity needs? Furthermore, Variable Length Subnet Masking (VLSM) allows you to use different subnet masks for each subnet, effectively "subnetting a subnet" to minimize wasted addresses. How do you prioritize your allocation steps to ensure that the largest host requirements are met without fragmenting your available IP space?