Expected Performance?

[weserhart]

Hi,

I am wondering what my expected performance should be.  I have a tunnel setup the Dallas server.

I am encountering high latency to ipv6 sites.  ipv4 latency seems to be around 30-70 msec but ipv6 latency is 130+msec.  Is this reasonable?  Does this indicate anything is wrong with my configuration?

Here some example pings:
[root@bigbird ~]# ping6 ipv6.google.com
PING ipv6.google.com(2001:4860:0:2001::68) 56 data bytes
64 bytes from 2001:4860:0:2001::68: icmp_seq=1 ttl=59 time=136 ms
64 bytes from 2001:4860:0:2001::68: icmp_seq=2 ttl=59 time=137 ms
64 bytes from 2001:4860:0:2001::68: icmp_seq=3 ttl=59 time=137 ms
64 bytes from 2001:4860:0:2001::68: icmp_seq=4 ttl=59 time=134 ms
^C
--- ipv6.google.com ping statistics ---

[root@bigbird ~]# ping google.com
PING google.com (64.233.187.99) 56(84) bytes of data.
64 bytes from jc-in-f99.google.com (64.233.187.99): icmp_seq=1 ttl=242 time=36.8 ms
64 bytes from jc-in-f99.google.com (64.233.187.99): icmp_seq=2 ttl=242 time=39.6 ms
64 bytes from jc-in-f99.google.com (64.233.187.99): icmp_seq=3 ttl=242 time=37.4 ms
64 bytes from jc-in-f99.google.com (64.233.187.99): icmp_seq=4 ttl=242 time=38.4 ms
64 bytes from jc-in-f99.google.com (64.233.187.99): icmp_seq=5 ttl=242 time=56.0 ms
^C
--- google.com ping statistics ---

[piojan]

First of all check the ping to tunnel server or have a look at the route to it

Code Select Expand

traceroute tserv8.dal1.ipv6.he.net
PJ

[weserhart]

PJ

I get long ping times (e.g. 60 msec).  The traceroute looks like there is something weird going on at step 10.

Code Select Expand


[root@bigbird ~]# traceroute tserv8.dal1.ipv6.he.net
traceroute to tserv8.dal1.ipv6.he.net (216.218.224.42), 30 hops max, 60 byte packets
1  x.x.net (192.168.1.1)  0.349 ms  0.630 ms  0.690 ms
2  cpe-76-185-48-1.tx.res.rr.com (76.185.48.1)  20.703 ms  20.711 ms  20.792 ms
3  gig1-4.dllatxpln-rtr2.tx.rr.com (70.125.216.45)  27.124 ms * *
4  gig5-0-0.dllatxchn-rtr5.tx.rr.com (70.125.217.104)  27.513 ms  27.501 ms  27.590 ms
5  gig0-1-0.hstntxl3-rtr1.texas.rr.com (72.179.205.74)  32.998 ms  33.109 ms  33.252 ms
6  xe-10-3-0.bar1.Houston1.Level3.net (4.79.88.21)  33.063 ms  30.592 ms xe-9-1-0.bar1.Houston1.Level3.net (4.79.88.25)  30.285 ms
7  ae-13-13.ebr1.Dallas1.Level3.net (4.69.137.138)  36.009 ms  24.929 ms  24.916 ms
8  ae-61-61.csw1.Dallas1.Level3.net (4.69.136.122)  27.540 ms ae-91-91.csw4.Dallas1.Level3.net (4.69.136.134)  23.930 ms ae-71-71.csw2.Dallas1.Level3.net (4.69.136.126)  34.433 ms
9  ae-43-99.car3.Dallas1.Level3.net (4.68.19.197)  25.875 ms ae-33-89.car3.Dallas1.Level3.net (4.68.19.133)  26.021 ms ae-43-99.car3.Dallas1.Level3.net (4.68.19.197)  25.848 ms
10  4.68.63.194 (4.68.63.194)  67.091 ms  66.996 ms  66.954 ms
11  las-bb1-link.telia.net (213.248.80.13)  67.148 ms  67.118 ms  67.267 ms
12  sjo-bb1-link.telia.net (213.248.80.18)  67.234 ms  67.201 ms  77.533 ms
13  hurricane-113209-sjo-bb1.c.telia.net (213.248.86.54)  84.576 ms  84.574 ms  84.684 ms
14  10gigabitethernet1-1.core1.lax1.he.net (72.52.92.22)  87.979 ms  87.524 ms  74.193 ms
15  10gigabitethernet1-2.core1.dal1.he.net (72.52.92.58)  71.309 ms  71.498 ms  71.280 ms
16  tserv8.dal1.ipv6.he.net (216.218.224.42)  71.187 ms  71.117 ms  76.947 ms


[snarked]

Well, your "long" times are because your IPv4 path to the tunnel server includes a round-trip to California instead of staying solely in Texas.  Either RR or Level3 needs better peering in state.

[piojan]

Either RR or Level3 needs better peering in state.

Mayby it's better now - traceroute from dallas to the 2nd hope (76.185.48.1):

Code Select Expand


  1    <1 ms   <1 ms   <1 ms TenGigabitEthernet1-4.ar3.DAL2.gblx.net [64.212.32.181]
  2    <1 ms   <1 ms   <1 ms 66.109.9.213
  3     3 ms   <1 ms   <1 ms ae-1-0.cr0.dfw10.tbone.rr.com [66.109.6.178]
  4    <1 ms   <1 ms   <1 ms 66.109.6.89
  5     4 ms   <1 ms   <1 ms gig2-0-0.dllatxchn-rtr5.tx.rr.com [72.179.205.73]
  6     5 ms    5 ms    5 ms 70.125.217.79
  7     6 ms    6 ms   *     gig7-2.dllatxpln-t-rtr1.tx.rr.com [70.125.217.105]
  8     6 ms    6 ms    6 ms cpe-76-185-48-1.tx.res.rr.com [76.185.48.1]


[malfeasant]

I think this shows the importance of doing a traceroute before selecting your tunnel server, geography can in some cases have nothing to do with how far your packets travel...
I'm in Phoenix AZ- I work only 5 miles from home, but when I shell into my home box (cox cable) from work (qwest dsl), my packets go to L.A. and back- in this case I would almost assume it's a no-brainer to pick the L.A. endpoint, but only almost, i plan to check latencies when i get home in a little while... in your case, given that your isp's "closest" connection to hurricane happens in california, you'd be better off with the L.A. or maybe even fremont server (i'm assuming "sjo-..." refers to san jose?)
I used to have a roadrunner cable connection many years ago, and it seemed they always had less than ideal routing, but who am I to criticise :D  I know a bit about NAT "routing", but next to nothing about the real thing...