You might want to consider SNMP monitoring the switch...pop it up in something like MRTG and you should see pretty quickly how often you are overwhelming your ports...

/nebulus