dslreports logo
 
    All Forums Hot Topics Gallery
spc
Search similar:


uniqs
494
slrdc
join:2001-01-09
Washington, DC

slrdc

Member

[Connectivity] Non-optimal IP Routing from Washington DC

Washington DC customer here and I've noticed that my data to other networks that used to go to an exchange point in Ashburn VA are now going all the way up to NYC before getting down there. Anyone know happened to the more direct link from DC to the exchange in VA and when this inefficiency started?


traceroute to www.nih.gov (137.187.25.43), 64 hops max, 52 byte packets
3 xe-10-2-0-32767-sur02.floridaave.dc.bad.comcast.net (68.87.132.233) 13.916 ms 30.975 ms 113.917 ms
4 68.85.115.197 (68.85.115.197) 22.879 ms 70.705 ms 26.957 ms
5 he-3-14-0-0-cr01.newyork.ny.ibone.comcast.net (68.86.94.101) 22.965 ms
he-3-4-0-0-cr01.newyork.ny.ibone.comcast.net (68.86.91.33) 33.446 ms
he-3-2-0-0-cr01.newyork.ny.ibone.comcast.net (68.86.94.177) 25.164 ms
6 be-10102-cr02.ashburn.va.ibone.comcast.net (68.86.85.25) 34.814 ms 34.238 ms 22.345 ms
7 be-10001-cr01.ashburn.va.ibone.comcast.net (68.86.85.1) 19.498 ms 66.213 ms 24.902 ms
8 he-0-10-0-0-pe07.ashburn.va.ibone.comcast.net (68.86.83.66) 20.116 ms 25.881 ms 33. 887 ms
9 ash1-pr1-ae5-301.us.twtelecom.net (66.192.252.209) 34.759 ms 52.620 ms 39.566 ms
10 ash1-ar4-xe-2-1-0-0.us.twtelecom.net (66.192.243.122) 51.966 ms 19.126 ms 26.643 ms
11 66-192-62-14.static.twtelecom.net (66.192.62.14) 27.742 ms 58.962 ms 21.630 ms
...

Napsterbater
Meh
MVM
join:2002-12-28
Milledgeville, GA

Napsterbater

MVM

[Connectivity] Re: Non-optimal IP Routing from Washington DC

Change in peering, Old route taken out of service, Capacity issues.

There can be many many many reasons, and nobody here can probably tell you why.
Kearnstd
Space Elf
Premium Member
join:2002-01-22
Mullica Hill, NJ

Kearnstd to slrdc

Premium Member

to slrdc
is it really less efficient, As in are speeds effected?

As for the routing itself maybe there is a problem on the more "direct" route.

NathanO
join:2008-08-21
Seattle, WA

NathanO to slrdc

Member

to slrdc
It looks like there were some network changes Yesterday/Today in your area. I'd expect this would be resolved shortly.

whfsdude
Premium Member
join:2003-04-05
Washington, DC

whfsdude to slrdc

Premium Member

to slrdc

Re: [Connectivity] Non-optimal IP Routing from Washington DC

Haven't looked on the legacy protocol yet. Did take a look on v6.

IPv6 routing is taking two different paths. Ignore the bad Chicago PTR.

traceroute to nihprod.cit.nih.gov (2607:f220:404:1001::89bb:192b) from 2001:559:800c:1020:225:90ff:fec2:beef, 30 hops max, 24 byte packets
 1  2001:559:800c:1020::1 (2001:559:800c::1)  0.258 ms  0.262 ms  0.177 ms
 2  2001:559:0:84::1 (2001:559:0:84::1)  0.437 ms  0.439 ms  0.37 ms
 3  ae-19-0-ar04.capitolhghts.md.bad.comcast.net (2001:558:340:27a::2)  0.869 ms  0.931 ms  0.927 ms
 4  he-2-0-0-0-cr01.chicago.il.ibone.comcast.net (2001:558:0:f6e2::1)  4.618 ms  2.954 ms  4.663 ms
 5  be-10001-cr01.ashburn.va.ibone.comcast.net (2001:558:0:f540::1)  5.459 ms  4.873 ms  4.052 ms
 6  he-0-11-0-0-pe07.ashburn.va.ibone.comcast.net (2001:558:0:f8d2::2)  2.626 ms  2.565 ms  2.512 ms
 7  ashburn-twt.host.twtelecom.net (2001:4870:4500:20::1)  2.162 ms  2.2 ms  2.161 ms
 8  2001:4871:ffff:f000:168:215:53:190 (2001:4871:ffff:f000:168:215:53:190)  26.961 ms  2.742 ms  2.88 ms
 9  2001:4870:4002:3::2 (2001:4870:4002:3::2)  4.19 ms  5.054 ms  4.82 ms
 

traceroute to nihprod.cit.nih.gov (2607:f220:404:1001::89bb:192b) from 2001:559:800c:1020:225:90ff:fec2:beef, 30 hops max, 24 byte packets
 1  2001:559:800c:1020::1 (2001:559:800c:1020::1)  0.282 ms  0.23 ms  0.206 ms
 2  2001:559:0:84::1 (2001:559:0:84::1)  0.452 ms  0.444 ms  0.383 ms
 3  ae-19-0-ar04.whitemarsh.md.bad.comcast.net (2001:558:340:27e::2)  3.176 ms  3.196 ms  3.153 ms
 4  he-3-4-0-0-11-cr01.newyork.ny.ibone.comcast.net (2001:558:0:f7ac::1)  7.928 ms  10.455 ms  7.948 ms
 5  be-10102-cr02.ashburn.va.ibone.comcast.net (2001:558:0:f546::1)  9.632 ms  10.274 ms  8.359 ms
 6  be-10001-cr01.ashburn.va.ibone.comcast.net (2001:558:0:f540::1)  9.755 ms  10.402 ms  7.897 ms
 7  he-0-11-0-0-pe07.ashburn.va.ibone.comcast.net (2001:558:0:f8d2::2)  7.907 ms  7.906 ms *
 8  ashburn-twt.host.twtelecom.net (2001:4870:4500:20::1)  7.541 ms  19.791 ms  7.506 ms
 9  2001:4871:ffff:f000:168:215:53:190 (2001:4871:ffff:f000:168:215:53:190)  8.216 ms  8.212 ms  8.256 ms
10  2001:4870:4002:3::2 (2001:4870:4002:3::2)  10.217 ms  9.614 ms  9.743 ms
 

Somewhat unrelated, nih.gov is not honoring path mtu discovery. Looks like they have a lower MTU (1460) and are filtering ICMP type 2 (packet too big).

# ping6 -s 1453 2607:f220:404:1001::89bb:192b
PING 2607:f220:404:1001::89bb:192b(2607:f220:404:1001::89bb:192b) 1453 data bytes
^C
--- 2607:f220:404:1001::89bb:192b ping statistics ---
29 packets transmitted, 0 received, 100% packet loss, time 28223ms
 

~# ping6 -s 1453 2001:4870:4002:3::2
PING 2001:4870:4002:3::2(2001:4870:4002:3::2) 1453 data bytes
^C
--- 2001:4870:4002:3::2 ping statistics ---
2 packets transmitted, 0 received, 100% packet loss, time 1008ms
 

# ping6 -s 1452 2001:4870:4002:3::2
PING 2001:4870:4002:3::2(2001:4870:4002:3::2) 1452 data bytes
1460 bytes from 2001:4870:4002:3::2: icmp_seq=1 ttl=54 time=4.65 ms
1460 bytes from 2001:4870:4002:3::2: icmp_seq=2 ttl=54 time=4.42 ms
^C
--- 2001:4870:4002:3::2 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1001ms
rtt min/avg/max/mdev = 4.422/4.536/4.650/0.114 ms
 
slrdc
join:2001-01-09
Washington, DC

1 edit

slrdc to Kearnstd

Member

to Kearnstd

Re: [Connectivity] Re: Non-optimal IP Routing from Washington DC

said by Kearnstd:

is it really less efficient, As in are speeds effected?

Latency is definitely affected. I do agree that there could be problems on the more "direct" route and I hope they get fixed!
slrdc

slrdc to whfsdude

Member

to whfsdude

Re: [Connectivity] Non-optimal IP Routing from Washington DC

said by whfsdude:

IPv6 routing is taking two different paths. Ignore the bad Chicago PTR.

How do you know it's a bad Chicago PTR in your hop #4?

My hop #5 (which is analogous to your hop #4) has all kinds of cities that it supposedly bounces through (not just NYC) before returning to Ashburn. I've seen Chicago, TX, and even _LA_. I also have IPv6 and I've seen the same things happen too (but mainly through NY).

I'd love to attribute all this to bad PTR records, but there seem to be too many of them and I think that it would drive the network engineers crazy if all of them were this wrong for this long.

whfsdude
Premium Member
join:2003-04-05
Washington, DC

whfsdude

Premium Member

The original routing issue (NYC) looks valid.
said by slrdc:

How do you know it's a bad Chicago PTR in your hop #4?

The bad DNS record on my first traceroute is ~4ms away.

From DC:
# ping6 -c 2 he-2-0-0-0-cr01.chicago.il.ibone.comcast.net
PING he-2-0-0-0-cr01.chicago.il.ibone.comcast.net(he-2-0-0-0-cr01.chicago.il.ibone.comcast.net) 56 data bytes
64 bytes from he-2-0-0-0-cr01.chicago.il.ibone.comcast.net: icmp_seq=1 ttl=60 time=4.55 ms
64 bytes from he-2-0-0-0-cr01.chicago.il.ibone.comcast.net: icmp_seq=2 ttl=60 time=3.06 ms
 

Assuming ~600 miles to Chicago from DC via a direct path (unlikely) Round trip is ~1200 miles. Fiber is about 125 miles per ms.

We end up at about 10ms as the minimum latency for a trip to Chicago if there was a very favorable path.