Are you using UDP or ICMP as your packet type? Is it just a single router and all downstream hops are working more like your actual performance might suggest?
Some routers have challenges properly routing ICMP "streams" of outgoing and incoming ICMP requests / replies. This can often be noticed at or near a NAT point, and can be related to the operation of doing the NAT address conversions. We've often heard of router bios updates fixing this.
PingPlotter uses ICMP TTL expired packets to measure the latency of all intermediate hops. The Linux (and Windows) PING utilities, though, use ICMP echo replies. PingPlotter, in ICMP mode, expects an ICMP echo reply from only the final destination. This could explain why no amount of ping traffic can cause this same behavior. You might be able to duplicate some of this by using the traceroute command (which uses similar packet types as PingPlotter in UDP mode) multiple times on the same workstation.
If your router is having problems returning ICMP TTL Expired packets to the right destination when it has numerous concurrent outstanding request from different program instances on the same PC (how's that for a list of conditions?), then try switching to UDP or updating your router bios.
If that doesn't make any difference, contact us at
support@pingplotter.com for some possible alternatives that are a bit more "product futures" related.
- Pete