Speeds.. (ntt.net)

[dataless]

First I'd like to say a huge thanks to you for your FREE services..

I have a question though, as I'm not a routing expert by any means..

I setup a tunnel to your Dallas endpoint because I'm in Houston.  The speeds ok but I'm wondering if they could be improved.  Below is a traceroute from the network where my server is located (Cbeyond T1) to your Dallas endpoint.

traceroute to 216.218.224.42 (216.218.224.42), 30 hops max, 40 byte packets
1  xx.xx.xx.xx (xx.xx.xx.xx)  7.606 ms  9.116 ms  10.047 ms
2  172.22.220.145 (172.22.220.145)  4.954 ms  5.282 ms  5.871 ms
3  192.168.39.22 (192.168.39.22)  8.234 ms  8.176 ms  8.467 ms
4  192.168.38.34 (192.168.38.34)  9.169 ms  9.110 ms  9.250 ms
5  ge-1-6.r03.hstntx01.us.bb.gin.ntt.net (128.241.1.177)  10.026 ms  10.508 ms  10.548 ms
6  xe-0-1-0.r20.hstntx01.us.bb.gin.ntt.net (129.250.2.228)  12.066 ms  8.477 ms  9.299 ms
7  p64-1-3-0.r20.dllstx09.us.bb.gin.ntt.net (129.250.3.129)  12.914 ms  12.510 ms  12.748 ms
8  po-2.r03.dllstx09.us.bb.gin.ntt.net (129.250.4.38)  12.193 ms  12.236 ms  12.724 ms
9  xe-0.globalcrossing.dllstx09.us.bb.gin.ntt.net (129.250.8.190)  14.886 ms  14.948 ms  14.956 ms
10  Hurrican-Electric-LLC.TenGigabitEthernet1-4.ar2.SJC2.gblx.net (64.214.174.246)  59.563 ms  59.805 ms  60.298 ms
11  10gigabitethernet3-2.core1.pao1.he.net (72.52.92.69)  69.398 ms  69.571 ms  69.898 ms
12  10gigabitethernet1-1.core1.lax1.he.net (72.52.92.22)  61.920 ms  58.122 ms  57.808 ms
13  10gigabitethernet1-2.core1.dal1.he.net (72.52.92.58)  55.101 ms  55.394 ms  67.907 ms
14  tserv8.dal1.ipv6.he.net (216.218.224.42)  67.468 ms  64.463 ms  63.750 ms

The slowdown seems to come between hops 9 and 10, going from what I assume is an ntt.net router in Dallas (dllstx09?) to your network uplink.  I'm not sure where ar2.sjc2.gblx.net is located though.

So I've got Houston -> Dallas -> ? -> LAX -> Dallas...

Is this just an issue between nnt.net and gblx.net that I can't do anything about I guess?

Your Ashburn endpoint actually seems slightly faster for me (about 5-8ms) but I have the exact same issue with the exact same companies..

traceroute to 216.66.22.2 (216.66.22.2), 30 hops max, 40 byte packets
1  xx.xx.xx.xx (xx.xx.xx.xx)  6.885 ms  8.192 ms  9.044 ms
2  172.22.220.145 (172.22.220.145)  5.315 ms  5.902 ms  5.981 ms
3  car01-s5-0-5.hstntx08.cbeyond.net (192.168.39.22)  8.527 ms  8.816 ms  9.093 ms
4  192.168.38.34 (192.168.38.34)  9.711 ms  10.006 ms  10.351 ms
5  ge-1-6.r03.hstntx01.us.bb.gin.ntt.net (128.241.1.177)  83.186 ms  83.531 ms  83.814 ms
6  xe-0-1-0.r20.hstntx01.us.bb.gin.ntt.net (129.250.2.228)  12.543 ms  8.278 ms  8.580 ms
7  p64-1-3-0.r20.dllstx09.us.bb.gin.ntt.net (129.250.3.129)  13.705 ms  13.686 ms  13.083 ms
8  po-2.r03.dllstx09.us.bb.gin.ntt.net (129.250.4.38)  12.078 ms  12.886 ms  12.959 ms
9  xe-0.globalcrossing.dllstx09.us.bb.gin.ntt.net (129.250.8.190)  13.400 ms  13.804 ms  14.154 ms
10  HURRICANE-ELECTRIC-LLC-Ashburn.TenGigabitEthernet4-4.ar3.DCA3.gblx.net (64.214.121.170)  58.780 ms  59.591 ms  59.671 ms

For some reason traceroute doesn't finish, but mtr works fine on this IP and shows less than a 1ms jump between dca3 and tserv13.ash1.ipv6.he.net.

Anyway, again an issue between ntt.net and gblx.net it looks like?


-------------- Added --------------

Just because I was curious, I figured I'd check from my home (Comcast) connection and see the path..

Tracing route to tserv8.dal1.ipv6.he.net [216.218.224.42]
over a maximum of 30 hops:

 1    <1 ms    <1 ms    <1 ms  10.0.10.1
 2     *        *        *     Request timed out.
 3     7 ms     8 ms     7 ms  [xx.xx.xx.xx]
 4     9 ms    10 ms     9 ms  te-9-4-ar01.royalton.tx.houston.comcast.net [68.85.244.253]
 5    10 ms     9 ms     9 ms  po-11-ar02.royalton.tx.houston.comcast.net [68.85.244.98]
 6    12 ms    13 ms    11 ms  po-17-ar02.greenspoint.tx.houston.comcast.net [68.85.244.130]
 7    18 ms    17 ms    18 ms  te-0-1-0-0-cr01.dallas.tx.ibone.comcast.net [68.86.91.49]
 8    19 ms    20 ms    19 ms  TenGigabitEthernet6-2.ar3.DAL2.gblx.net [67.17.157.221]
 9    61 ms    64 ms    63 ms  Hurrican-Electric-LLC.TenGigabitEthernet1-4.ar2.SJC2.gblx.net [64.214.174.246]
10    64 ms    63 ms    73 ms  10gigabitethernet3-2.core1.pao1.he.net [72.52.92.69]
11    62 ms    64 ms    63 ms  10gigabitethernet1-1.core1.lax1.he.net [72.52.92.22]
12    62 ms    61 ms    61 ms  10gigabitethernet1-2.core1.dal1.he.net [72.52.92.58]
13    62 ms    61 ms    61 ms  tserv8.dal1.ipv6.he.net [216.218.224.42]

Trace complete.

Same result with Hurrican-Electric-LLC.TenGigabitEthernet1-4.ar2.SJC2.gblx.net from another network.

[snarked]

What this is saying is that NTT peers with Global Crossing (gblx) in Dallas, Texas but NTT doesn't peer with HE at Dallas (or at all).  Gblx peers with HE, but in San Jose, California.  That's why the traffic is going to and from Dallas to and from San Jose.  If HE peered with NTT or gblx at Dallas, your traffic latency would be solved.  Although both have a presence in Dallas, what we don't know is whether the peering points that currently exist are at the same or at different facilities.  If they're at the same place, then it may be just a matter of stringing a cable between their routers.

[dataless]

Since I know nobody at either company to even contact and ask about peering, I guess I'll wait and see if they eventually work something out as HE's tunnels become more and more popular.

Thanks for the translation of the issue.

Again, for a FREE service that I'm using to experiment with and serve up some HTTP and that's about it, 150ms isn't a massive deal.

I must say HE's setup has been very easy to work with and I have learned a lot about IPv6 in the past week.

[dataless]

According to peeringdb.com both HE and Comcast share many Equinix locations in common, it seems though Comcast is picky about peering.  http://www.comcast.com/peering/

I didn't find any US information about NTT Communications Corporation (ntt.net) on peeringdb.com though.


Just researching because I'm curious of the odds they would peer sometime down the road.