Smokeping

Here is a great tool that I’ve used over the years to help troubleshoot ISP latency issues and QoS issues when working with VoIP lines, but it can be used to troubleshoot all sorts of issues: Smokeping

You can use this wonderful tool at DSL Reports: http://www.dslreports.com/smokeping

What does SmokePing do?

SmokePing generates flexible graphs that, within hours, contain actual information about the quality & reachability of your IP address from several distributed locations.

Why is low latency and zero packet loss so important?

Communications over the internet are built on layers of technology, and one of the lowest and most fundamental layers is how quickly and reliably IP data packets are transmitted and received. If this layer breaks down then almost everything else is affected in one way or another: Speed drops, web pages mysteriously hang, programs disconnected unexpectedly, VOIP quality suffers .. and so on.

Ok I’m sold, what do I do?

If you want to do a quick check of the quality of an IP address, then simply elect to have smokeping monitor it closely for 24 hours, (or more, if you are a registered user of the site).

The main result page for an IP you are monitoring shows an overview graph for each monitoring station. At the moment, there are three stations. The red line on the graph represents the average ping time over the time period. The graphs scroll to the left, with new data appearing on the right.

If you click any individual monitoring station (click the graph or the navigation links to the left), you are able to see the ping plot in more detail. In particular, the color of the line segments (color key at the bottom of the graph) indicates overall packet loss. “Smoke” indicates variability of latency (ping time) at that instant. The more variability, the more “smoke” that appears around the colored line segment.

Because our implementation of SmokePing uses LARGE packet sizes, and each data point on the graph is not just a single measurement, but many pings, the graph is more sensitive than normal ping plotting programs and will uncover latency variability or loss that may only show up when the line is under fairly heavy use.

The three monitoring stations are independent from each other so they will also confirm a problem as yours if they all show the same suspicious data (excessive smoke, a rise in average ping time or actual packet loss) at the same time. Of course, if just ONE out of three monitoring stations shows a problem, then you can probably discount it.

What should I hope to see?

Here is an example of a nice stable connection NEAR to a monitoring station. Although there is “smoke,” there is no packet loss, and the smoke is only +/- a few milliseconds. (Click thumbnail for full size.)

Here is an example of the same connection from the other side of the country. Notice that since the ping time is much higher, the variability (smoke) is almost invisible. (Click thumbnail for full size.)

When viewing the results, please keep in mind that your IP may be very near to one of the stations and much further away from the other two. The graphs auto-scale, so “smoke” shown on an X scale composed of just a few ms top to bottom is far less interesting than smoke shown on a scale of 100ms! Loss, (line segments that are not green) on any scale, is always interesting!

Please visit this page later for some examples of both good and bad SmokePing 24 hour result graphs!

You can also visit the SmokePing home page where there is another short description of how to interpret SmokePing data: reading smokeping graphs.