10

I'm getting increased latency and StDev due to route congestion and packet loss, but the forward and reverse paths go over distinct networks (e.g. one being init7.net, the other one being he.net), so, it's very difficult to understand which network or host is responsible for the congestion, packet loss, jitter and increased latency.

Is there a way to narrow down the blame after forward and reverse mtr fails to pinpoint the exact culprit, and the NOC@ contacts either don't respond, or claim to suffer no loss on the path in question? (I'm using OpenBSD.)

I've even tried doing an mtr directly to some customers of both of the two networks that might be experiencing the congestion, but couldn't really find any problems that way, especially since, for example, he.net has many POPs, and oftentimes distinct routes are taken between a given entry and exit POP, so when I try to mtr their hosts (like the tserv) directly at the exit POP to which I might be losing packets in their network, a different he.net path is taken to reach the very same POP, and no packet loss occurs, which proves nothing of interest (other than a possible suggestion that they might indeed overload some routes, whilst ensuring others stay uncongested, all whilst ignoring NOC@ requests from non-customers).

  • Did any answer help you? if so, you should accept the answer so that the question doesn't keep popping up forever, looking for an answer. Alternatively, you could provide and accept your own answer. – Ron Maupin Aug 8 '17 at 15:15
9

One way to do this is ICMP Timestamp, which is milliseconds from midnight UTC. It has the added benefit that you don't necessarily need to control both ends, as long as the far-end is not firewalled, there is good chance it'll work.

However, to have reliable one-way measurements, you need reliably same time in both ends. As ICMP timestamp only have precision of 1ms (which is not nearly enough for many applications, but sufficient for this) it's reasonably easy to find even non-cooperating hosts where ICMP timestamp will provide useful data.

If you control both ends, be sure that you are synchronizing NTP to only 1 server and same server. The absolute clock is not very important, it's just important that you experience as closely same time as possible.

If ICMP timestamp is not sufficient, it's very easy to write 10 lines of ruby/perl/python or even C to do measurements when you control both ends.

I can't really suggest software for doing ICMP timestamp measurements unidirectionally, hping2 supports sending ICMP timestamp but for some reason does not output unidirectional values. I wrote patch for hping2 to display one way latencies.

  • Wow, your hping --icmp-ts extra arithmetic is so bloody awesome! Too lazy to get the sources and recompile the binary, I got myself a shell version of your hping patch (stackoverflow.com/q/20172028/1122270), and it shows a pretty much constant time over the init7 path to hetzner, and an all-over-the-map variance with the he.net path from hetzner! I finally have definitive proof that init7 is telling the truth! Although I'd argue against using the same ntp server just for this: just make sure an ntpd is alive (I didn't have to change any settings at all, yet values look reasonable). – cnst Nov 24 '13 at 8:50
  • 1
    If you care about accurate wall time, you'll want at least 3 NTP servers (To be able to detect false ticker, 2 NTP servers is worst option you could do). But here we don't care about wall time, we care about ticking accurately at same clock, regardless of wall. So for 1-way measurement optimal results come from accurate clock, wall time being irrelevant. Happy to hear you got results! – ytti Nov 24 '13 at 9:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.