All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Telia poor routing to Dacentec... Ongoing issue?
pubcrawler
Banned
Anyone else noticing goofy non direct routing to Dacentec in Lenoir, NC?
Our provider (Time Warner) routes everything out through Chicago. From there, providers tend to pickup traffic with direct routes (major peering point).
With Telia heading to Dacentec the route is all messed up.
Chicago to Dallas to Atlanta to Lenoir
10: network-065-029-001-028.midwest.rr.com 31.875ms
11: ae-4-0.cr0.chi30.tbone.rr.com 33.208ms
12: 107.14.17.147 69.291ms
13: chi-bb1-link.telia.net 41.365ms
14: chi-bb1-link.telia.net 34.303ms asymm 15
15: dls-bb1-link.telia.net 51.691ms asymm 14
16: atl-bb1-link.telia.net 53.014ms asymm 15
17: dacentec-ic-154273-atl-bb1.c.telia.net 77.176ms asymm 16
18: dct-cr01--v51.dacentec.com 92.212ms asymm 17
19: dct-ds04--vl56.dacentec.com 88.930ms asymm 18
20: 199.241.189.66 77.082ms asymm 19
21: 199.241.189.68 92.328ms reached
Ping times to Dacentec in Lenoir as-is are rather high for us, 70-90ms.
What should be a long driving distance of 500 miles ends up being many times more through this wild round about routing. Chicago to Dallas driving miles is 926. Dallas to Lenoir is driving miles 1028 =2000 miles?!?!
Even if we take a more wise route of Chicago to Buffalo to Lenoir = 534 driving miles. Buffalo to Lenoir = 641 driving miles. Total 1200 miles.
Obviously, I need to monitor these routes and do more testing. Telia is strong if you catch it through Cleveland to Buffalo to Ashburn (see this route most commonly via Colocrossing affiliated). But if you get anywhere near Chicago we hit that common corridor down to Dallas and over to Atlanta which is too darn far and wrong direction.
Comments
Cost drives routing too much. Its a terrible pattern in the industry.
Raise the issue with your carrier, if the network admins are kind they might actually look at it, otherwise file it in the special bin shaped drawer.
I've seen this routing geography field trip before. HE I see in doing the same Chicago to Dallas weirdness.
Whoever lays these networks out needs to refer to a fullsized wall map before hard coding such geographic preferences.
90ms is high latency. That's a US east to west coast sort of time and that is even highly inflated.
It's not that they aren't thinking about a destination, just often not the same one you are. Adding routes costs, when the right pieces don't fall into place, has to be good incentive. My take on it anyway. I'm no expert
Could be worse though...still not fdc
Unsure why we all continue to think Chicago is geographically the center of the country. It's a good 400-500 miles from due center I'd estimate. It is also too far north to be literal center.
Chicago is on the west side of the US East looking at a map I'd say. Doesn't US Central Time Zone begin near there?
Dallas in contrast is nearer to due center, but very distant from Chicago.
As for FDC, their route to Dacentec looks stellar in contrast:
2 te0-5-0-0.ccr22.ord03.atlas.cogentco.com (38.104.103.45) 0.778 ms 0.774 ms 0.839 ms
3 te0-4-0-0.ccr21.ord01.atlas.cogentco.com (154.54.43.229) 0.816 ms te0-4-0-0.ccr22.ord01.atlas.cogentco.com (154.54.1.97) 0.811 ms te0-3-0-5.ccr21.ord01.atlas.cogentco.com (154.54.1.1) 0.870 ms
4 te0-1-0-5.ccr22.atl01.atlas.cogentco.com (154.54.28.74) 17.838 ms te0-2-0-4.ccr22.atl01.atlas.cogentco.com (154.54.29.113) 17.793 ms 17.795 ms
5 te3-1.ccr01.clt01.atlas.cogentco.com (154.54.6.174) 22.653 ms 22.666 ms te2-1.ccr01.clt01.atlas.cogentco.com (154.54.0.97) 22.706 ms
6 dacentec.demarc.cogentco.com (38.104.168.10) 25.416 ms 25.418 ms 25.444 ms
7 dct-ds04--vl56.dacentec.com (199.255.156.198) 25.509 ms 25.515 ms 25.593 ms
8 199.241.189.66 (199.241.189.66) 25.568 ms 25.546 ms 25.460 ms
9 199.241.189.68 (199.241.189.68) 25.555 ms 25.558 ms 25.477 ms
25ms from Chicago to Lenoir?!?! compared to 90ms over Telia...
Odd as a 25 hour day, because Dacentec has upstream from Cogent per their public peering info via HE.
Looking at the Telia network map, it looks like it can either run from Chicago to Dallas to Atlanta or from Chicago to NYC to DC to Atlanta. They don't have a Chicago to Atlanta direct route.
@qps, couldn't get Telia's flash-based map to work. Do you have a image file of the map?
Also couldn't get HE's Looking Glass to load, as HE is another upstream and I expect to take similar long field trip but want to confirm.
Just filed a comprehensive ticket with Dacentec about the routing. We'll see how they do cleaning that up. I'd be glad to see packets flow over Cogent instead.
HE's routing between Chicago and Lenoir is equally round about and high latency for the distance between the two end points:
1 26 ms 24 ms 28 ms 10gigabitethernet1-2.core1.ash1.he.net (184.105.213.246)
2 33 ms 37 ms 50 ms 10gigabitethernet1-2.core1.atl1.he.net (184.105.213.110)
3 28 ms 28 ms 27 ms dacentec.tieatl.telxgroup.net (198.32.132.98)
4 70 ms 71 ms 74 ms dct-cr01--v51.dacentec.com (199.255.156.53)
5 73 ms 76 ms 73 ms dct-ds04--vl56.dacentec.com (199.255.156.198)
6 75 ms 74 ms 74 ms 199.241.189.66
Just sent it to you by e-mail.
Thanks @qps.
Odd long route there on Telia from Dallas to NYC directly it looks. Suspect they have hops along the way. Nothing to cross over and interconnect from Chicago though to Atlanta. So true about the Chicago to Cleveland to Buffalo to NYC to Ashburn to Atlanta versus the Chicago to Dallas to Atlanta.
HE routes look saturated from Ashburn, VA to Lenoir:
1 23 ms 13 ms 23 ms 10gigabitethernet1-2.core1.atl1.he.net (184.105.213.110)
2 15 ms 21 ms 12 ms dacentec.tieatl.telxgroup.net (198.32.132.98)
3 86 ms 75 ms 74 ms dct-cr01--v51.dacentec.com (199.255.156.53)
4 74 ms 74 ms 75 ms dct-ds08-vl-41 (199.241.190.230)
5 72 ms 71 ms 71 ms 192.198.8...
i opeened a ticket and they said they let telia know.
Thanks @fly. More tickets mean better priority.
I have high hopes for Dacentec. Former foray into North Carolina with another provider ended in flames and disappointment.
Seeing the route to Dacentec from our location remaining the same, so far.
However, the latency has come down some. Instead of seeing 90ms often earlier, are down to 59ms on the low side, 64ms average.
Makes me think there was a flood style attack earlier or just prime time latency increase.
59ms is still to high, but an improvement.
going over telia is fixed now. 60ms compared to 160ms
Tried it it was bad so didn't collide there also this is the reason there cheap.
let's everyone say it together:
IT'S NOT DACENTEC'S PROBLEM, IT'S TELIA UPSTREAM
also, the high latency is back
Either way, It doesn't sop the data center from being crap. Its in a disaster zone, in NC "Why hosts in NC?" And they say dedicated 1Gbit port but its highly seems as its shared between the whole rack, as I got 1 to test, and that's the last test im doing.
Sounds like you had an unusually bad first experience. Sure it's not the best around but there's plenty that are equal to or worse that are highly utilized in this market. It's pretty darn nice for what it is. A lot of LEB companies can't survive on much higher quality because they either can't profit on it or they end up overselling themselves to death to do it.
52ms from Chicago to Lenoir via Telia's own looking glass. That's better than yesterday.
25.2ms solid over Cogent from same geographic end points.
Lenoir might appear in a red circle for earthquake risks, but it isn't earthquake central. Closest quake was 22 miles away from Lenoir.
See: http://www.homefacts.com/earthquakes/North-Carolina/Caldwell-County/Lenoir.html
As far as the gigabit per rack, that would be mighty shady if that's what they are doing. I have no proof of such. But, their sales specials clearly say for portions of racks that 100Mbps on Gbit port. That indicates a single ethernet run. This is fairly common and especially at bargain rates.
See: http://www.webhostingtalk.com/showthread.php?t=1218137
I don't see why you would expect more than 1 gig port unless you were paying for that on top of plan for more bandwidth.
how is NC in a disaster zone? what disaster? no earthquakes, don't think tornadoes, hurricanes once every like 50 years. if you think NC is in a "disaster zone", you're clearly mistaken.
I think you should stop posting
you're probably just mad cuz you fucked something up and support wouldn't hold your hand
Dacentec staff / support really dropping in my opinion.
My ticket about the latency was ended here:
Mitchell T Staff 01/22/2013 01:10:04
Hello,
We have opened a support ticket with the provider having the issues(TeliaSonera) and we are awaiting their response.
Thanks,
Mithcell
Now, the ticket has been closed. Attempted to re-open it to determine what the issue was/is and resolution.
Nope, I can't re-open the ticket. WTF?
They probably didn't focus on the auto-close function.
you should just be able to respond to the ticket even if it's "closed"
also, @pubcrawler you should consider trying to directly contact telia
it seems to be an intermittent issue, as I'm not seeing any latency increases atm
I wouldn't say NC is a disaster zone, but I also wouldn't say it's not impossible for a tornado or torrential rain from a hurricane to cause flooding that may damage a data center. From a quick google, it looks like NC has recently had major tornados sweep through.
http://en.wikipedia.org/wiki/April_14–16,_2011_tornado_outbreak
There is no "perfect" place for a data center, and you really have to compromise on what you can afford and what kind of risk your willing to take with your project/business/etc.
If it helps anyone I am going to post some random traceroutes I did to the same dacentec ip (one of my servers). I will first put the one from my RR home account which uses telia also. I LIVE in NC and I have the highest ping of all the other locations.
1 172.16.1.1 (172.16.1.1) 1.085 ms 0.920 ms 3.286 ms
2 * * *
3 gi15-0-0-3102.gnboncsg-rtr2.triad.rr.com (24.28.228.65) 12.436 ms 21.907 ms 11.478 ms
4 ge-3-1-0.chrlncpop-rtr1.southeast.rr.com (24.93.64.162) 13.772 ms 57.963 ms 12.544 ms
5 107.14.19.46 (107.14.19.46) 20.826 ms
107.14.19.18 (107.14.19.18) 18.645 ms
107.14.19.48 (107.14.19.48) 17.440 ms
6 107.14.19.11 (107.14.19.11) 18.845 ms
ae-0-0.pr0.atl20.tbone.rr.com (66.109.6.171) 18.010 ms 17.662 ms
7 atl-bb1-link.telia.net (80.239.160.61) 21.803 ms 17.200 ms 17.960 ms
8 dacentec-ic-154273-atl-bb1.c.telia.net (213.155.129.114) 34.058 ms 32.780 ms 33.455 ms
9 dct-cr01--v51.dacentec.com (199.255.156.53) 48.282 ms 49.924 ms 49.603 ms
10 dct-ds08-vl-41 (199.241.190.230) 43.334 ms 44.720 ms 53.289 ms
11 (192.198.) 43.665 ms !Z 45.573 ms !Z 40.875 ms !Z
traceroute to (192.198.), 30 hops max, 60 byte packets
1 .static.hostnoc.net () 1.045 ms 1.405 ms 1.863 ms
2 ec0-58.agg03.sctn01.hostnoc.net (64.120.243.69) 6.531 ms 6.533 ms 6.529 ms
3 xe1-04.gwy02.sctn01.hostnoc.net (96.9.191.5) 0.456 ms 0.949 ms 0.951 ms
4 TenGE2-1.br02.phl02.pccwbtn.net (63.218.31.41) 4.986 ms 5.503 ms 5.502 ms
5 cogent.tenge13-4.br02.ash01.pccwbtn.net (63.218.92.14) 136.810 ms 136.813 ms 136.808 ms
6 te0-1-0-6.mpd21.iad02.atlas.cogentco.com (154.54.31.161) 9.221 ms 9.380 ms 9.740 ms
7 te0-5-0-2.mpd21.dca01.atlas.cogentco.com (154.54.1.205) 9.268 ms te0-5-0-5.mpd21.dca01.atlas.cogentco.com (154.54.1.229) 9.252 ms te0-5-0-2.mpd21.dca01.atlas.cogentco.com (154.54.1.205) 9.380 ms
8 te4-4.ccr01.clt01.atlas.cogentco.com (154.54.84.254) 106.955 ms 106.959 ms 106.953 ms
9 dacentec.demarc.cogentco.com (38.104.168.10) 27.397 ms 27.378 ms 27.900 ms
10 dct-ds08-vl-41 (199.241.190.230) 27.863 ms 27.870 ms 27.862 ms
11 (192.198.) 27.473 ms !X 27.448 ms !X 27.470 ms !X
traceroute to (192.198.), 30 hops max, 60 byte packets
1 .securedragon.net () 0.077 ms 0.017 ms 0.014 ms
2 208.38.174.218 (208.38.174.218) 0.867 ms 0.919 ms 1.040 ms
3 gi2-4.border4.esnet.com (216.139.207.25) 0.716 ms 0.755 ms 0.802 ms
4 208.67.164.221 (208.67.164.221) 4.863 ms 5.152 ms 5.399 ms
5 208.67.167.50 (208.67.167.50) 13.404 ms 13.447 ms 13.431 ms
6 208.67.164.89 (208.67.164.89) 9.635 ms 9.174 ms 9.243 ms
7 208.67.165.101 (208.67.165.101) 14.622 ms 15.040 ms 15.414 ms
8 208.67.164.5 (208.67.164.5) 26.548 ms 26.531 ms 26.513 ms
9 dacentec.tieatl.telxgroup.net (198.32.132.98) 14.383 ms 14.370 ms 14.350 ms
10 dct-cr01--v51.dacentec.com (199.255.156.53) 24.765 ms 24.752 ms 24.814 ms
11 dct-ds08-vl-41 (199.241.190.230) 25.075 ms 25.061 ms 25.075 ms
12 (192.198.) 24.839 ms !X 24.824 ms !X 24.878 ms !X
traceroute to (192.198.), 30 hops max, 60 byte packets
1 199.168. (199.168.) 0.489 ms 1.028 ms 1.515 ms
2 206.131.207.125 (206.131.207.125) 0.158 ms 0.261 ms 0.327 ms
3 69.30.209.137 (69.30.209.137) 0.367 ms 0.444 ms 0.500 ms
4 10gigabitethernet1-3.core1.mci2.he.net (216.66.78.89) 0.417 ms 0.460 ms 0.545 ms
5 10gigabitethernet1-1.core1.mci3.he.net (184.105.213.38) 0.423 ms 0.490 ms 0.548 ms
6 10gigabitethernet4-4.core1.dal1.he.net (184.105.222.10) 13.660 ms 13.710 ms 13.719 ms
7 10gigabitethernet5-4.core1.atl1.he.net (184.105.213.114) 34.549 ms 33.890 ms 33.979 ms
8 dacentec.tieatl.telxgroup.net (198.32.132.98) 33.917 ms 34.024 ms 34.119 ms
9 dct-cr01--v51.dacentec.com (199.255.156.53) 39.771 ms 39.822 ms 39.893 ms
10 dct-ds08-vl-41 (199.241.190.230) 40.029 ms 40.120 ms 40.176 ms
11 (192.198.) 39.834 ms !X 39.828 ms !X 39.852 ms !X
Yikes @Holoshed. That latency over Time Warner and Telia is horrendous.
40ms+ for a same state end to end.
Unsure what Telia's problem is, but something is clearly not as it should be.
Watch how the cogent path bypasses 199.255.156.53 entirely... That's your clue
Actually looking at that now, I see what you mean. That one point adds almost double to the time.
Wondering if the 199.255.156.53 is in Atlanta at peering point or if the IP is on the other side after the peering point.
Suspect it is on the opposing side of the peering at the facility.
The would point to there being clog/saturation of the connection between the facility and peering. That would typically be an end-facility purchased amount hitting the ceiling.
Problem is that IP isn't always showing big increase in latency.
Finding some threads about folks howling at the moon about Telia connectivity leaving Chicago. So appears Telia is somewhat know in some places for latency ballooning. First I've seen of it though.
Then again, a few months ago I posted on here about why is everyone going Telia crazy latey (seeing it from many providers).
perhaps someone is filling the pipe at that pop? maybe getting dos'd or something. Or that pipe's just oversold, because I noticed latency spikes at night, but not as much during the day.