|
Slow IPv6 since World IPv6 day?
![[gb]](/s/countries/gb.gif) Shadow Hawkins on Monday, 27 June 2011 23:07:25
Hello all,
I wonder if anyone else is having slow IPv6 connectivity since around the time of World IPv6 day?
Some times it is fine, but it seems that later in the day, it gets very slow.
I am on gblon03, and although my v4 ping to the POP (212.113.147.150) is decent,
10 packets transmitted, 10 received, 0% packet loss, time 9014ms
rtt min/avg/max/mdev = 35.065/41.815/72.048/11.153 ms
my ping times to the other end of my v6 tunnel are bad:
29 packets transmitted, 25 received, 13% packet loss, time 28028ms
rtt min/avg/max/mdev = 214.606/241.666/341.694/38.629 ms
I don't know of any change that can cause this - I use AICCU (as this is a laptop, and I don't always know or control what network device I am behind) and Fedora 14.
The only thing I can think of is that since World IPv6 day, some big sites have left v6 enabled, and a lot more traffic is going over the PoP's networks, which is causing delays?
My Sixxs tunnel graph shows long ping times, and packet loss too.
I have only 2 UK ISPs I can test with, so I was wondering if anyone else is having any problems like this?
Calum
Slow IPv6 since World IPv6 day?
The PoPs are not doing that more traffic as before, the only thing that is causing a few more bits is Youtube though.
As for your pings, they are nice, but they don't show where this is happening nor over which path, and you left out the source and destination addresses too which make the numbers quite useless.
As such, a traceroute over v4 from your endpoint to the PoP and a ping from that endpoint to the PoP would be useful.
And I do hope that you are not testing this on wireless or behind some ISP which ratelimits...
Slow IPv6 since World IPv6 day?
Shadow Hawkins on Tuesday, 28 June 2011 20:40:44
Hello Jeroen,
I didn't bother with traceroute, as I was pinging the IPv4 endpoint (which was fast), and the ::1 part of my tunnel.
Here is the traceroute though:
$ traceroute6 2a00:14f0:e000:2e::1
traceroute to 2a00:14f0:e000:2e::1 (2a00:14f0:e000:2e::1), 30 hops max, 80 byte packets
1 gw-47.lon-03.gb.sixxs.net (2a00:14f0:e000:2e::1) 518.612 ms 520.221 ms 537.994 ms
$
$ traceroute 212.113.147.150
traceroute to 212.113.147.150 (212.113.147.150), 30 hops max, 60 byte packets
1 xxx (192.168.1.1) 1.617 ms 1.838 ms 2.012 ms
2 88-111-xx-xx.dynamic.dsl.as9105.com (88.111.xx.xx) 37.020 ms 38.347 ms 41.242 ms
3 xe-4-0-0.ner001.the.as13285.net (80.40.155.102) 45.587 ms 47.216 ms 49.095 ms
4 80.40.155.107 (80.40.155.107) 88.069 ms 80.40.155.103 (80.40.155.103) 52.861 ms 80.40.155.111 (80.40.155.111) 87.475 ms
5 xe-8-3-0.bragg001.log.as13285.net (80.40.155.33) 56.620 ms xe-9-3-0.bragg002.log.as13285.net (80.40.155.41) 58.358 ms 60.180 ms
6 xe-5-3-0.scr001.log.as13285.net (80.40.155.50) 63.207 ms 39.576 ms 40.632 ms
7 host-78-144-0-168.as13285.net (78.144.0.168) 64.199 ms host-78-144-0-172.as13285.net (78.144.0.172) 41.398 ms 44.105 ms
8 xe-0-0-3.border-1.sov.lon.uk.as29017.net (195.66.224.141) 44.232 ms 44.354 ms 45.435 ms
9 ae4.core-1.lhc.lon.uk.as29017.net (89.145.125.17) 47.874 ms 49.687 ms 50.837 ms
10 gblon03.sixxs.net (212.113.147.150) 53.510 ms 55.688 ms 56.817 ms
$
For some reason, the v6 is very slow, when the v4 is fine.
I am on wireless (on this particular tunnel), but this hasn't affected things in the past.
As I say, this is a new thing, and it only affects v6 - my v4 ping times to the endpoint are nice and low (30-40ms), which is fine for the setup here.
I'm not sure about the ISP, which is why I asked on the forum, to see if it was a wider problem. I don't know if the ISP ratelimits - do we have a list?
Calum
Slow IPv6 since World IPv6 day?
Seems that at least once they did
http://www.raymond.cc/blog/archives/2010/07/21/glasnost-tells-if-your-isp-is-limiting-torrent-speed-in-8-minutes/
http://www.the-scream.co.uk/forums/t18124.html
A quick google reveals that they still might be doing it.
I would not be surprised at all if they are throttling 'strange' ports. AYIYA they would then definitely be strange and classified as something they don't like.
Slow IPv6 since World IPv6 day?
![[us]](/s/countries/us.gif) Shadow Hawkins on Wednesday, 29 June 2011 02:09:57
Greetings,
I noticed this thread and it seemed somewhat similar to my circumstance.
Background: My connectivity is residential ADSL with Verizon,
current v4 IP is 71.96.137.134 (dynamic assignment),
location Plano, Texas,
POP is usdal01 with v4 IP address of 209.197.16.66,
tunnel id is T72013 and is AYIYA via AICCU from FreeBSD 7.4 ports tree.
I received a tunnel grant approximately 1 week ago and have been watching the
latency graph for my tunnel via the tunnel detail information page of Sixxs
(thanks Sixxs). For the last few days, I have noticed that my general
latency seems to be around 30 ms, but there is a daily spike of duration
2-3 hours that goes to a max of over 100 ms. The spike begins rising at
1700 CDT and peaks around 2200 CDT. (I use local time since you will see I
think the delays are local.) So, following suggestions
from Jeroen Massar earlier in this thread, I decided to investigate.
I have collected a few samples of IPv4 traceroutes at various times.
Here is a sample from low latency (approx 15:25 CDT on 2011 June 28):
Traceroute to (209.197.16.66)
1 L100.DLLSTX-DSL-06.verizon-gni.net (71.96.128.1) 22 ms 22 ms 20 ms
2 P2-1.DLLSTX-LCR-02.verizon-gni.net (130.81.34.66) 22 ms 22 ms 22 ms
3 so-5-0-0-0.DFW01-BB-RTR2.verizon-gni.net (130.81.29.178) 24 ms 22 ms 22 ms
4 0.ae2.BR1.DFW13.ALTER.NET (152.63.2.149) 26 ms 24 ms 26 ms
5 te9-1-10G.ar5.DAL2.gblx.net (64.208.110.205) 52 ms 26 ms 24 ms
6 po2-20G.ar2.DAL2.gblx.net (67.16.136.25) 26 ms 24 ms 26 ms
7 * (67.17.194.78) 24 ms 26 ms 26 ms
8 1-3.r1.da.hwng.net (209.197.8.9) 36 ms 28 ms 26 ms
9 * (209.197.16.66) 26 ms 28 ms 26 ms
Samples during high latency are consistent - here is 19:02 CDT on 2011 June 28
Traceroute to (209.197.16.66)
1 L100.DLLSTX-DSL-06.verizon-gni.net (71.96.128.1) 22 ms 22 ms 20 ms
2 P2-1.DLLSTX-LCR-02.verizon-gni.net (130.81.34.66) 24 ms 22 ms 22 ms
3 so-5-0-0-0.DFW01-BB-RTR2.verizon-gni.net (130.81.29.178) 22 ms 22 ms 24 ms
4 0.ae2.BR1.DFW13.ALTER.NET (152.63.2.149) 24 ms 24 ms 24 ms
5 te9-1-10G.ar5.DAL2.gblx.net (64.208.110.205) 24 ms 24 ms 26 ms
6 te3-3-10G.ar2.DAL2.gblx.net (67.16.132.193) 24 ms 24 ms 24 ms
7 * (67.17.194.78) 106 ms 108 ms 104 ms
8 1-3.r1.da.hwng.net (209.197.8.9) 158 ms 114 ms 134 ms
9 * (209.197.16.66) 138 ms 112 ms 112 ms
Note that the increased transit time seems to involve line 7 - the v4 IP
address is similar to lines 5 and 6 and dig -x returns no record but a SOA
record as follows:
;; AUTHORITY SECTION:
194.17.67.in-addr.arpa. 10800 IN SOA gblx.net. dns.gblx.net.
2011053100 7200 1800 604800 43200
based on WHOIS, this is Global Crossing Limited (now Level 3).
My conclusion is that my latency difficulties are on the IPv4 links for
the tunnel and not an IPv6 issue. Looks like gblx.net has some extra
traffic for a couple of hours at the end of business day here in Dallas.
Of course, all except the statistics are supposition on my part.
Regards to all,
Jerry Schieffer
|
![[ch]](/s/countries/ch.gif)
on Monday, 27 June 2011 23:49:41
Posting is only allowed when you are