We love it when a product keeps getting better. We don't mean products that just lose a few bugs. We mean products that get new and exciting features that really improve the functionality even as the bugs are eradicated. Such is the case with a product that has been a longtime favorite of ours: PingPlotter from Nessoft.

PingPlotter, which runs under Windows, has been invaluable around here. We have found so many opportunities to use it since we first reviewed it back in 1 BC (that's "Beyond Crash," or what the rest of the world calls 2001) that it would be hard to imagine not having it around.

Should you not have total recall of that column or are disinclined to explore the dust-laden Gearhead catacombs , let us briefly recap how this tool works: PingPlotter is essentially good ol' *nix traceroute with a pretty face and muscles.

The original traceroute program was the usual "as friendly as a cornered rat" text program, whereas PingPlotter is graphical and user friendly. With this release, Version 2.7, PingPlotter has become traceroute with a pretty face and muscles on amphetamines and drinking Red Bull.

PingPlotter is primarily used to continuously track the routes to servers, but the latest version (only $20 - the first version is available as freeware) takes the product to new levels of functionality.

The original traceroute program determines the sequence of routers that lie between the machine running traceroute and some target machine, and reports on the ping time (the latency, or round-trip duration) to each router.

The way traceroute works is cunning: It first finds the target machine's IP address and then sends the target an Internet Control Message Protocol (ICMP) request with a Time to Live (TTL) parameter set to 1.

The first router receives the packet and decrements the TTL value by one, finds it is 0, and responds with a TTL-exceeded response that includes the router's IP address, which we can resolve into a name with a reverse DNS lookup. And since we know when we sent the packet and when the response from the router was received we could determine the latency.

Now that we know the first step in the chain of routers, we can send an ICMP message with TTL set to 2 and determine the IP address, name and latency of the next router and so on for each router hop in turn.