Large Subnets and Discovery
An address range that include more than 2000 nodes takes much longer (one to two hours, for example) to discover than the same number of nodes split up into multiple smaller ranges. Additionally, with so many nodes on a map, the user interface and NTM operations may run with noticeable lag.
For example, if you are subnetting with the mask of 255.255.248.0, then the maximum number of nodes within the subnet will be 8 X 255 = 2040. In discovery nodes, the software engine creates a lookup table in memory that includes as many rows as nodes in the defined IP range or subnet. The more rows the more time the engine must spend in finding its point of reference in the table as it iterates through the array of items. Walking a larger lookup table takes significantly more time than walking smaller tables that cumulatively contain the same number of arrayed items. So the time it takes the engine to complete its discovery task directly depends on the number of possible nodes in the specified range or subnet.