| carrier - cable Is start.ca using only HE as their carrier for cable services?
In every traceroute i only notice HE.
If this is the case, why? Do you have any backup carrier?
Do you expect to get more carries for better routing?
Thanks |
|
 roccaStart.caPremium
join:2008-11-16
London, ON
kudos:11 | We use Bell, HE, Cogent and are peered with TorIX and other various interconnections. |
|
roccaStart.caPremium
join:2008-11-16
London, ON
kudos:11 | reply to JohnTO
»bgp.he.net/AS40788 has lots of interesting information if you're interested... |
|
Reviews:
 ·TekSavvy DSL
2 edits | reply to JohnTO
I too have noticed only HE is being used for transit for outbound routes, even when it appears Cogent would provide a better route . I have run traceroutes to all of the monitoring stations here: »www.nimsoft.com/content/Nimsoft/···ons.html. For all except one (where Torix is used) HE is used on the outbound.
For inbound my IP is announced through HE and Cogent (no Bell) (»route.robtex.com/108.170.128.0-1···-ca.html) and the inbound routes appear to be an equal mix between HE and Cogent. But on the outbound HE appears to be heavily favoured to the point of being almost exclusively used. |
|
roccaStart.caPremium
join:2008-11-16
London, ON
kudos:11 | Roughly, Hurricane takes 45%, Bell 25%, Cogent 20% and TorIX/other peers about 10% |
|
Reviews:
·TekSavvy DSL
1 edit | said by rocca:Roughly, Hurricane takes 45%, Bell 25%, Cogent 20% and TorIX/other peers about 10%
That may be the case but Cogent and Bell are not represented at all in outbound routes from my connection.
Here are some examples where HE is being used even though Cogent would provide a better route.
Here is a route to a Nuclear Fallout (Internap) VPS in Chicago:
traceroute to 64.74.97.93 (64.74.97.93), 64 hops max, 72 byte packets
1 10.204.100.1 (10.204.100.1) 8.668 ms 7.517 ms 8.455 ms
2 asr-link1-burlington1.net.start.ca (64.140.112.154) 19.310 ms 19.861 ms 19.518 ms
3 core1-toronto1-te3-3.net.start.ca (64.140.112.153) 22.083 ms 22.962 ms 19.875 ms
4 v521.core1.tor1.he.net (209.51.164.81) 27.162 ms 24.609 ms 26.826 ms
5 10gigabitethernet7-3.core1.chi1.he.net (184.105.213.150) 35.550 ms 36.411 ms 34.658 ms
6 64.62.180.94 (64.62.180.94) 35.694 ms 37.173 ms 36.228 ms
7 207.88.15.2.ptr.us.xo.net (207.88.15.2) 33.018 ms 33.676 ms 33.660 ms
8 207.88.184.146.ptr.us.xo.net (207.88.184.146) 36.897 ms 35.452 ms 53.853 ms
9 border5.po2-bbnet2.chg.pnap.net (64.94.32.74) 35.916 ms 42.818 ms 34.924 ms
10 c-64-74-97-93.managed-vds.internap-chicago.nfoservers.com (64.74.97.93) 35.732 ms 34.964 ms 38.314 ms
I know the VPS is announced through Cogent and this is the route I took when with Cogeco:
traceroute to 64.74.97.93 (64.74.97.93), 64 hops max, 72 byte packets
1 10.90.192.1 (10.90.192.1) 8.588 ms 8.393 ms 8.540 ms
2 d226-12-93.home.cgocable.net (24.226.12.93) 21.851 ms 22.138 ms 20.384 ms
3 d226-6-134.home.cgocable.net (24.226.6.134) 23.297 ms 22.936 ms 22.940 ms
4 te4-1.ccr01.yhm01.atlas.cogentco.com (38.122.17.17) 31.427 ms 30.960 ms 29.811 ms
5 te3-7.ccr01.buf02.atlas.cogentco.com (154.54.82.101) 30.881 ms 31.566 ms 29.737 ms
6 te8-8.ccr02.cle04.atlas.cogentco.com (154.54.31.237) 27.396 ms 28.039 ms 28.097 ms
7 te0-5-0-6.ccr21.ord01.atlas.cogentco.com (154.54.45.130) 31.052 ms 28.801 ms 28.944 ms
8 te0-1-0-3.ccr22.ord03.atlas.cogentco.com (154.54.1.2) 27.830 ms 27.695 ms 27.981 ms
9 38.104.102.102 (38.104.102.102) 31.270 ms 39.784 ms 32.979 ms
10 border5.po2-bbnet2.chg.pnap.net (64.94.32.74) 32.220 ms 32.015 ms 32.238 ms
11 c-64-74-97-93.managed-vds.internap-chicago.nfoservers.com (64.74.97.93) 29.387 ms 29.678 ms 27.653 ms
For whatever reason, even though Cogent is supposed to be used as one of the transit provides my route goes Start --> HE --> XO --> Internap, rather than Start --> Cogent --> Internap. Latency is the same either way but the current route includes an extra AS hop.
Another example. The New York Watchmouse monitor station:
traceroute to ny.watchmouse.net (208.85.4.114), 64 hops max, 72 byte packets
1 10.204.100.1 (10.204.100.1) 11.359 ms 6.992 ms 7.726 ms
2 asr-link1-burlington1.net.start.ca (64.140.112.154) 20.298 ms 21.351 ms 20.122 ms
3 core1-toronto1-te3-3.net.start.ca (64.140.112.153) 23.077 ms 19.277 ms 19.593 ms
4 v521.core1.tor1.he.net (209.51.164.81) 20.595 ms 23.729 ms 24.789 ms
5 10gigabitethernet7-3.core1.chi1.he.net (184.105.213.150) 35.877 ms 33.777 ms 40.270 ms
6 eqix.chcg.twtelecom.net (206.223.119.36) 33.580 ms 33.668 ms 32.887 ms
7 aby1-ar1-xe-0-0-0-0.us.twtelecom.net (66.192.242.50) 46.925 ms 43.587 ms 57.243 ms
8 66.192.207.178 (66.192.207.178) 41.786 ms 43.017 ms 43.749 ms
9 66-206-32-189.static.turnkeyinternet.net (66.206.32.189) 38.931 ms 40.417 ms 39.200 ms
10 usnyc01.watchmouse.net (208.85.4.114) 40.319 ms 40.352 ms 41.084 ms
This route also goes out via HE despite being reachable more directly by Cogent. See my route when with Cogeco.
traceroute to ny.watchmouse.net (208.85.4.114), 64 hops max, 72 byte packets
1 10.90.192.1 (10.90.192.1) 6.962 ms 8.783 ms 7.616 ms
2 d226-12-85.home.cgocable.net (24.226.12.85) 23.067 ms 23.656 ms 22.519 ms
3 d226-6-134.home.cgocable.net (24.226.6.134) 20.981 ms 22.095 ms 19.662 ms
4 te4-1.ccr01.yhm01.atlas.cogentco.com (38.122.17.17) 30.560 ms 30.990 ms 30.786 ms
5 te7-7.ccr01.buf02.atlas.cogentco.com (154.54.24.221) 30.153 ms 29.736 ms 30.124 ms
6 te8-7.ccr02.alb02.atlas.cogentco.com (154.54.27.177) 30.308 ms 31.539 ms 29.599 ms
7 38.104.52.58 (38.104.52.58) 33.225 ms 38.813 ms 31.337 ms
8 usnyc01.watchmouse.net (208.85.4.114) 31.864 ms 31.103 ms 32.154 ms
Current route goes Start --> HE --> Time Warner Telecom --> Turnkey, rather than Start --> Cogent --> Turnkey
So for whatever reason HE is being heavily favoured and used exclusively on the outbound (from my connection anyway). |
|
roccaStart.caPremium
join:2008-11-16
London, ON
kudos:11 | said by vonSchroeder:So for whatever reason HE is being heavily favoured and used exclusively on the outbound (from my connection anyway).
There are lots of metrics at play, but yes not a lot of paths will transit Cogent for off-net routes. We're also adding another large player in November and routing policy may change again then based on real-world patterns. We continuously monitor, measure and report to ensure we are providing the best performance and reliability. |
|
