CentOS 5 Networking :: ICMP Ping Latency Grows Over Time

I have a dedicated host on my lan to monitor other hosts/services using Nagios. I'm in the process of migrating to Zabbix on that host to perform the same purpose. Both Nagios and Zabbix monitor icmp ping latency (Nagios uses ping, Zabbix uses fping) and over time the latency to other hosts grows until threshold alarms are triggered. In one week, the average latency grows from sub-millisecond to over 100 milliseconds, and continues to grow until the Nagios host is rebooted. I have verified the latency numbers using ping/fping from the command line on the Nagios host.

The problem is that pings from the monitored hosts to the Nagios host show normal latency at the time the Nagios host is showing high latency from itself to the monitored hosts. The Nagios host and monitored hosts are all connected to the same Dell 24 port gigabit switch. I already posted this question on the Zabbix forums with a graph of the latency but there were no answers. [URL] why the icmp ping latency is growing over time and how I can fix it short of rebooting the host on a schedule.

[Code]...