The last important timing improvement flag to understand is delaying and increasing rates directly. This is probably the most fine-tuned of the performance optimization, and should generally only be used if there is a specific problem you are trying to solve, or a situation you are trying to create.

The first of these flags, --scan-delay, specifies the amount of time that Nmap should wait, not doing anything, between probes. This can be extremely useful in slowing scans down (and sometimes speeding them up, with a low scan-delay). Again, the most common usage of slowing down scans is to avoid detection of a target administrator or security system, or to try to avoid network throttling issues. Since many systems use the rate at which requests are being made to determine whether or not the machine is under attack, this can be a very stealthy technique without very much effort on the part of the tester. It's also useful to note that --max-scan-delay can be...