11-06-2017 07:59 PM - edited 11-06-2017 08:11 PM
(Edit: I had a brain fail and read YYC as Toronto instead of Calgary, but have updated the post now)
Not sure who best to feed this to, needs to end up with the Telus NOC, to improve their IPv6 routing
On the Vancouver IPv6 network, the IPv6 routing to the Canadian BCNET/CANARIE networks is really terrible.
It goes Vancouver, Seattle, Vancouver, Calgary, Vancouver!
Telus/Hurricane Electric Hops:
Vancouver -> Seattle
Hurricane Electric Hops
Seattle -> Vancouver -> Calgary
CANARIE/BCNET Hops:
Calgary -> Vancouver
Telus should be peering at the Vancouver Internet Exchange (http://www.vanix.ca/participants/), or at the very least, Peer with Hurricane Electric in Vancouver, rather than Seattle.
Separately, I'm going to contact BCNET & Hurricane Electric to figure out why the traffic over the HE network went to Calgary and back.
Redacted traceroutes:
Telus -> BCNET
traceroute to 2001:410:1001:1::2 (2001:410:1001:1::2), 30 hops max, 80 byte packets 1 node-1w7jr9qttjiq1nlioXXXXXX.ipv6.telus.net (2001:569:7ca6:XXXX::1) 0.309 ms 0.272 ms 0.252 ms 2 node-1w7jr9n36nixrvaceajc2psux.ipv6.telus.net (2001:569:2:f::59) 3.399 ms 11.957 ms * 3 sttlwawbbr03.bb.telus.com (2001:568:1::50d) 4.937 ms 5.069 ms 5.049 ms 4 v6-10gigabitethernet1-3.core1.sea1.he.net (2001:504:16::1b1b) 5.742 ms 5.731 ms 5.714 ms 5 100ge10-1.core1.yvr1.he.net (2001:470:0:3ac::2) 11.768 ms 11.699 ms 11.695 ms 6 100ge10-2.core1.yyc1.he.net (2001:470:0:3ad::2) 18.572 ms 18.400 ms 18.134 ms 7 gw-canarie.yycix.ca (2001:504:2f::6509:1) 18.466 ms 19.427 ms 19.046 ms 8 vncv1rtr1.canarie.ca (2001:410:101:47::1) 29.841 ms 29.825 ms 30.148 ms 9 2001:410:101:b::2 (2001:410:101:b::2) 30.394 ms 30.366 ms 29.943 ms 10 2001:410:1001:1::2 (2001:410:1001:1::2) 30.200 ms 30.164 ms 30.156 ms
BCNET return path to Vancouver:
traceroute to 2001:569:7ca6:XXXX::1 (2001:569:7ca6:XXXX::1), 30 hops max, 80 byte packets 1 2607:f8f0:6a0:f001:de9f:dbff:fe16:646e (2607:f8f0:6a0:f001:de9f:dbff:fe16:646e) 0.526 ms 0.678 ms 0.914 ms 2 2607:f8f0:10:102::1 (2607:f8f0:10:102::1) 1.347 ms 1.397 ms 1.378 ms 3 2001:410:1001:1::1 (2001:410:1001:1::1) 1.415 ms 1.363 ms 1.331 ms 4 vncv1rtr1.canarie.ca (2001:410:101:b::1) 1.456 ms 1.428 ms 1.371 ms 5 clgr2rtr1.canarie.ca (2001:410:101:47::2) 12.246 ms 12.240 ms 12.242 ms 6 gw-hurricane-electric.yycix.ca (2001:504:2f::6939:1) 20.271 ms 19.330 ms 19.344 ms 7 100ge10-2.core1.yvr1.he.net (2001:470:0:3ad::1) 25.963 ms 24.897 ms 24.819 ms 8 100ge10-2.core1.sea1.he.net (2001:470:0:3ac::1) 25.578 ms 25.584 ms 25.615 ms 9 node-1w7jr9fenjgbk2vrqtjijbcxp.ipv6.telus.net (2001:568:1::13d) 25.931 ms 25.931 ms v6-six2.bb.telus.com (2001:504:16::433:0:354) 25.739 ms 10 node-1w7jr9n36nixrvaceajc2psux.ipv6.telus.net (2001:569:2:f::59) 29.963 ms 29.647 ms 29.464 ms 11 node-1w7jr9qttjiq1nliosXXXXX.ipv6.telus.net (2001:569:7ca6:XXXX::1) 30.032 ms 30.169 ms 30.017 ms _
11-06-2017 08:06 PM
YYC is Calgary, not Toronto.
11-06-2017 08:10 PM
Brainfart, but it's still going further than it should.
11-07-2017 09:34 AM
Hey Rob
What do you mean by bad IPv6 routing? Are you referring to the peering arrangement or packet delivery latency? These are not necessarily related.
I can certainly ask my NOC teams for input but at 30ms packet delivery latency, this is considered "very good" in the grand scheme of things.
Jordan
11-07-2017 12:33 PM
The latency is really high due to the routing via bad peering, the final destination for the data is in Vancouver.
It's funny I ran into this after my previous experiences with longer-distance IPv6 having better latency.
The IPv4 latency to the similar final systems is <10ms. IPv6 is 30ms because of the convoluted routing.
Telus's IPv4 traffic connects to Hurricane Electric in Vancouver.
Telus's IPv6 traffic connects to Hurricane Electric in Seattle but should connect in Vancouver (VANIX!)
Separately Hurricane Electric is sending BCNET-destined IPv6 traffic via Calgary first, and their IPv4 traffic is also staying local in Vancouver :-(. I have tickets separately with them to figure out that issue, but suspect it's policy related.
$ ping -i 0.1 -q -c 10 teleport.iat.sfu.ca PING sr-teleport.iat.sfu.ca (209.87.60.113) 56(84) bytes of data. --- sr-teleport.iat.sfu.ca ping statistics --- 10 packets transmitted, 10 received, 0% packet loss, time 905ms rtt min/avg/max/mdev = 5.438/6.183/6.588/0.302 ms $ traceroute teleport.iat.sfu.ca traceroute to teleport.iat.sfu.ca (209.87.60.113), 30 hops max, 60 byte packets 1 172.16.XXX.XXX (172.16.XXX.XXX) 0.136 ms 0.158 ms 0.196 ms 2 10.31.XXX.1 (10.31.XXX.XXX) 8.657 ms 17.823 ms 18.581 ms 3 154.11.10.11 (154.11.10.11) 4.501 ms ae0.100.STTLWAWBGR80.bb.telus.com (154.11.10.9) 4.449 ms 4.490 ms 4 GTD-HTRV-R1.bb.telus.com (154.11.15.10) 5.345 ms 5.405 ms 5.347 ms 5 100ge10-1.core1.yvr1.he.net (184.105.64.110) 8.357 ms 8.382 ms 8.191 ms 6 bcnet.10gigabitethernet1-4.core1.yvr1.he.net (184.105.148.150) 8.706 ms 7.474 ms 7.477 ms 7 cr1-tx3920.vantx2.BC.net (207.23.253.34) 4.567 ms 5.247 ms 5.252 ms 8 606-TX-SFU1-VNCV1-cr2.vncv1.bc.net (134.87.30.241) 10.723 ms 10.015 ms 10.219 ms 9 van-hcc1360-fw-1-van-hcc1360-border-1.net.sfu.ca (142.58.254.137) 6.509 ms 5.927 ms 4.989 ms 10 142.58.254.134 (142.58.254.134) 5.984 ms 5.998 ms 5.179 ms 11 van-hcc1360-core-1-van-hcc1360-fw-1.net.sfu.ca (142.58.254.129) 5.665 ms 8.932 ms 8.854 ms 12 sry-sc3685-x-1-van-hcc1360-x-1.net.sfu.ca (142.58.29.97) 6.665 ms 6.679 ms 6.378 ms 13 142.58.191.242 (142.58.191.242) 6.067 ms 6.916 ms 6.814 ms
(traceroute is firewalled past this point) 14 * * * 15 * * * 16 * * * 17 * * * 18 * * *
11-07-2017 12:39 PM
I will relay this to the appropriate NOC teams. Will not promise any guarantees but I do appreciate this constructive feedback.
Jordan
11-15-2017 03:38 PM
Hey
I have spoken with quite a few of our internal teams and unfortunately we cannot state publicly the current relationship status or future planning state with our global carriers. Thank you for your suggestion however.