Es gibt keine optimale MTU!
Warum eine feste MTU nicht „optimal“ sein wird.
niehaus@ceramic:~$ ping -c 1 -s 1465 -M do google.com
PING google.com (173.194.113.131) 1465(1493) bytes of data.
From ceramic.fritz.box (192.168.0.21) icmp\_seq=1 Frag needed and DF set (mtu = 1492)
--- google.com ping statistics ---
0 packets transmitted, 0 received, +1 errors
niehaus@ceramic:~$ ping -c 1 -s 1464 -M do google.com
PING google.com (173.194.113.131) 1464(1492) bytes of data.
1472 bytes from ham02s11-in-f3.1e100.net (173.194.113.131): icmp\_req=1 ttl=54 time=36.9 ms
--- google.com ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 36.940/36.940/36.940/0.000 ms
niehaus@ceramic:~$ ^C
niehaus@ceramic:~$ ping6 -c 1 -s 1464 -M do google.com
PING google.com(ham02s11-in-x03.1e100.net) 1464 data bytes
From ceramic-2.fritz.box icmp\_seq=1 Packet too big: mtu=1472
--- google.com ping statistics ---
0 packets transmitted, 0 received, +1 errors
niehaus@ceramic:~$
Okay?
niehaus@ceramic:~$ ping -c 1 -s 1464 -M do www.kame.net
PING orange.kame.net (203.178.141.194) 1464(1492) bytes of data.
--- orange.kame.net ping statistics ---
1 packets transmitted, 0 received, 100% packet loss, time 0ms
niehaus@ceramic:~$ ping -c 1 -s 146 -M do www.kame.net
PING orange.kame.net (203.178.141.194) 146(174) bytes of data.
154 bytes from orange.kame.net (203.178.141.194): icmp\_req=1 ttl=42 time=267 ms
--- orange.kame.net ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 267.840/267.840/267.840/0.000 ms
niehaus@ceramic:~$
Huch! Finger weg von den Filtern.
Auf der Suche nach der optimalen MTU:
niehaus@ceramic:~$ ping -c 1 -s 1453 -M do www.kame.net
PING orange.kame.net (203.178.141.194) 1453(1481) bytes of data.
^C
--- orange.kame.net ping statistics ---
1 packets transmitted, 0 received, 100% packet loss, time 0ms
niehaus@ceramic:~$ ping -c 1 -s 1452 -M do www.kame.net
PING orange.kame.net (203.178.141.194) 1452(1480) bytes of data.
1460 bytes from orange.kame.net (203.178.141.194): icmp\_req=1 ttl=42 time=283 ms
--- orange.kame.net ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 283.429/283.429/283.429/0.000 ms
Frage an die Experten: Wie lautet nun die optimale MTU?
…
…
…
…
niehaus@ceramic:~$ ping6 -c 1 -s 1425 -M do www.kame.net
PING www.kame.net(2001:200:dff:fff1:216:3eff:feb1:44d7) 1425 data bytes
From ceramic-2.fritz.box icmp\_seq=1 Packet too big: mtu=1472
--- www.kame.net ping statistics ---
0 packets transmitted, 0 received, +1 errors
niehaus@ceramic:~$ ping6 -c 1 -s 1424 -M do www.kame.net
PING www.kame.net(2001:200:dff:fff1:216:3eff:feb1:44d7) 1424 data bytes
1432 bytes from 2001:200:dff:fff1:216:3eff:feb1:44d7: icmp\_seq=1 ttl=48 time=319 ms
--- www.kame.net ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 319.973/319.973/319.973/0.000 ms
niehaus@ceramic:~$
Wie war noch gerade die optimale MTU?
niehaus@ceramic:~$ ping6 -c 1 -s 1500 -M want www.kame.net
PING www.kame.net(2001:200:dff:fff1:216:3eff:feb1:44d7) 1500 data bytes
1508 bytes from 2001:200:dff:fff1:216:3eff:feb1:44d7: icmp\_seq=1 ttl=48 time=328 ms
--- www.kame.net ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 328.683/328.683/328.683/0.000 ms
niehaus@ceramic:~$
Na geht doch!
Sebastian