Home / server / Questions / 306724
Accepted
Sander Steffann
Asked: 2011-09-01 00:48:57 +0800 CST

ping6 overhead compared to ping overhead

  • 772

I am testing IPv6 latency from a Linux box, and I noticed this weird difference between IPv4 ping and IPv6 ping:

# ping -n -A -q -c 500 speedtest.steffann.nl
PING speedtest.steffann.nl (10.3.10.20) 56(84) bytes of data.

--- speedtest.steffann.nl ping statistics ---
500 packets transmitted, 500 received, 0% packet loss, time 240ms
rtt min/avg/max/mdev = 0.297/0.364/7.213/0.317 ms, ipg/ewma 0.481/0.358 ms

Average rtt is 0.364, the count is 500, so that accounts for 182ms. The runtime of 240ms is a bit higher, but that is not a surprising amount of overhead. Now the IPv6 ping:

# ping6 -n -A -q -c 500 speedtest.steffann.nl
PING speedtest.steffann.nl(fd9c:262f:e839:310::20) 56 data bytes

--- speedtest.steffann.nl ping statistics ---
500 packets transmitted, 500 received, 0% packet loss, time 5000ms
rtt min/avg/max/mdev = 0.508/0.751/2.197/0.254 ms, pipe 2, ipg/ewma 10.021/0.725 ms

The rtt is about 2 times as long, so I would also expect a runtime of about 2 times as long. But it's more than 20 times as long. And it is exactly 10ms per ping...

It is probably an implementation artifact somewhere. Does anybody know where this comes from?

linux ping latency ipv6 linux-networking

1 Answers

  1. Best Answer

    Perhaps there is something funny going on with the IPv6 ping command. It sure sounds like '-A' is not doing what it is supposed to with ping6.

    Having said that ... IPv6 may not be as fast as IPv4, especially in the early phases of IPv6. There are just not as many organizations routing IPv6, fewer IPv6 peers. Some IPv6 packets may be going over tunnels. The path your IPv6 packet takes might be longer. A traceroute6 might show you that.

    I attempted to ping Google via IPv4 and IPv6 and I got nearly identical results.

    • 1