Hey
I'm seeing a funky route between AS5089 (Virgin Media) and yourselves, I've already shot them a message but was wondering if you could also look into it as well :)
alex@blackhole:~$ traceroute -A -q1 216.66.80.26
traceroute to 216.66.80.26 (216.66.80.26), 30 hops max, 60 byte packets
1 router.lan.alexsmith.im (192.168.1.254) [AS28513] 0.359 ms
2 10.240.72.1 (10.240.72.1) [AS65534] 17.309 ms
3 bmly-core-1b-ge-300-1167.network.virginmedia.net (81.100.0.205) [AS5089] 18.265 ms
4 brnt-bb-1b-as3-0.network.virginmedia.net (212.43.162.61) [AS5089] 19.233 ms
5 telc-ic-1-as0-0.network.virginmedia.net (62.253.185.74) [AS5089] 19.264 ms
6 ldn-b2-link.telia.net (213.248.100.97) [AS1299] 70.753 ms
7 ldn-bb1-link.telia.net (80.91.249.171) [AS1299] 69.758 ms
8 adm-bb3-link.telia.net (80.91.250.84) [AS1299] 79.334 ms
9 adm-b5-link.telia.net (80.91.247.215)
- 78.348 ms
10 hurricane-109025-adm-b5.c.telia.net (213.248.72.146) [AS1299] 80.248 ms
11 10gigabitethernet1-4.core1.lon1.he.net (72.52.92.81) [AS6939] 79.574 ms
12 tserv5.lon1.ipv6.he.net (216.66.80.26) [AS6939] 83.631 ms
It's perfectly fine from my own BGP network in London so I figured it must be a peering issues between your two ASN's.
Cheers,
Alex
Edit: It looks like they have issues with their AMS-IX connectivity but I'm pretty certain you used to peer with them at LINX?
I'm also on virgin media and have had the same issue for a few week, HE had a look and said it was a problem with the way virgin media routes the connection.
I tried talking to virgin media about it but trying to find someone that understands is mission impossible, i hope you have more luck than me.
at least its not as bad as it was a few weeks ago when they (virgin) were routing us via ash1.he.net then nyc4.he.net
then finally lon1.he.net
Correct, we do peer with them at LINX and the issue is their return path to us, they prefer to send traffic over Telia and back to us rather than hand it back to us over the peering session. I emailed them about this but didn't hear anything back. I'll try and poke them again.
Thanks. I've been trying to explain it to them as it's a wider problem that they are having (latency quadrupling in peak times isn't good), hopefully now I've armed myself with Smokeping graph's I'll have some more luck :) I just thought it was odd they didn't do it via LINX with their LINX peers, unless they don't have the spare capacity ???