This is not as easy to answer as it might seem o the face of it. First of all, are you asking about latency on a given interface or network segment, or latency to a particular host? Or are you interested in the throughput you can achieve on an interface or the throughput you can achieve to a particular host? Are you concerned about sheer number of packets / bits per second, or sustainable throughput over a TCP stream? E.G. I can flood a host w/ UDP at what ever wirespeed my host can support, but the infrastructure between points a & b might not support that speed. And if your concern is file transfers, there's TCP over head to take into consideration.

all that being said, look at iperf / jperf. They're a good pair of testing tools.

U can use command line tool like netstat for eg. type this in terminal..... netstat -tpe or one more is
tracert i.e. traceroute
or if ur router supports SNMP, u can use MRTG to monitor traffic on the router interfaces.
U can install smtpd on ur Linux (OS flavor dependent) gateway and use mrtg to monitor the gateway interfaces.

If u want to use graphical tools, u may try for Munin http://munin-monitoring.org/
or RRDTool: http://oss.oetiker.ch/rrdtool/