How to figure out source of internet latency



  • I've been having some issues with my internet connection. It's a local WISP that works fairly well most of the time and when it's not a phone call is usually all it takes to get it sorted out. Well things have changed and the regular tech support guy at my ISP moved on to greener pastures. Unfortunately there is no one worth a damn left to fill his shoes. They try, but they just don't have to knowledge that the previous guy did. Which brings me to today.

    I've been having issues with my connection since about 8PM las Saturday evening. It's been slow and unresponsive and the source is high sporadic latency. Ping times are running anywhere from 160 to 200+ ms. Throughput is there most of the time but the connection is flakey at best because of the latency.

    So far they say there is no problem on their end. They said today that ping time from tower to tower and to my antenna are ~8ms. I've checked everything here by pinging the gateway and wan from my desktop (0ms) and the problem is not with any of my equipment.

    What's the best way to go about figuring out where the latency is being introduced so I can point them in the right direction? Keep it simple I'm an average Joe not an IT guy.  :)



  • <sb>Holy crap! That sucks</sb>
    I'd start with a traceroute and watch the latency at each hop.
    Setting the gateway monitor to something like google dns helps with getting decent RRD quality data.



  • Traceroute did show some spikes in latency at various points but it's real inconsistent. But things have gotten worse in the last couple hours so I should have a better indication where the problem lies.

    Thanks for the tip on the alternate gateway monitoring function, I didn't even know it existed. Hopefully this will produce something to give my ISP an idea of what's going on.

    I can't help but think how bad this is that I have to provide my ISP with info they need, but should be able to produce, to track down a problem in their network.



  • Start a large download to your network then do the traceroute.

    Windows machine?  Ping each hop for 200-300 pings  C>Ping -n 300 (ip address)  and see where it gets squirrelly.



  • Traceroute is too limited!

    Pathping will show you latency across each hop!



  • There is a PFSense MTR package that is like traceroute but with configurable samples.



  • Try PingPlotter.
    Google DNS might have some latency problems on its own during "rush hours". I wouldn't specifically use them. See screenshot attached.




  • Well the issue seems to be resolved now. I launched a bunch of tracerts to see where the latency was happening and it's at my ISP's "end of the line" (their wording) where their network ends and is handed off to their provider.  I changed the monitoring to track the latency at the end hop IP. Showed quite a different picture once that was changed. Of course they still say it's not their problem and scheduled a tech to come out Tuesday to take a look at it which of course I'll be charged for if they can't find anything wrong. But since it's fixed now I'll cancel the appointment.

    I wish I had other options but I'm not willing to give up my rural location just for better internet. Thanks for all the suggestions and I'll keep them all in mind when this happens again.  :-\



  • Well looks like I spoke too soon.  :(



  • Run Ping Test. From the tools page, select Start, in the Ping Test (Real Time) box. This will advance you to a page indicating that all of the listed servers will be ping-ed twice per second and every thirty (30) seconds a report on your connection from A to F will be provided.


Log in to reply