If you suspect there is a latency issue, my first recommendation is to gets many, many samples. Network people can be extremely gruff. And if you crawled around on your hands and knees all day stringing cable, or spent hours staring at network graphs, you would probably be the same way.

Modern networks can be programmed to change packet routing based on packet type, time of day, and activity intensity. The only way to really understand latency issues is to take a lot of samples, paste them into Excel, and create a scatter graph.

It is important that your sample latency packets look just like SQL*Net packets. One of the best ways to do this is to use Oracle's tnsping command, found in the $ORACLE_HOME/bin directory.7 While both the ping and tnsping commands provide network latency times, the classic ping command is not good enough. Sharp network administrators managing a complex network infrastructure could possibly delay a standard ping packet, stop it altogether, or route it around the world a few times. So simply avoid this trap by issuing the tnsping command.