Checking Jumbo Frames setting on Linux

I've notice that transferring large files to newly installed FC6 machine was sluggish. I've been using different Linux distributions and Windows Server 2003 on that computer but transferring large files wasn't this sluggish. I don't know whether this is going to help but I'm enabling Jumbo Frames. I didn't noticed and speed improvement when I enabled on Windows server but hopefully it will help on Linux.

BTW, at least both machines and router/switch have to support Jumbo Frames.

• From the shell.

      #ifconfig eth0 mtu 9000

• Or you can manually enter the setting in /etc/init.d/net and restart the network.

• Or you can do it on GUI: System/Administration/Network then change the MTU value.

To check whether Jumbo Frames are working, ping the computer with Fragment disabled. When Jumbo Frames was disabled:

  c:\>ping -l 8500 -f 192.168.1.100
  Pinging 192.168.1.100 with 8500 bytes of data:
  Packet needs to be fragmented but DF set.
  Packet needs to be fragmented but DF set.
  Packet needs to be fragmented but DF set.
  Packet needs to be fragmented but DF set.
  Ping statistics for 192.168.1.100:
      Packets: Sent = 4, Received = 0, Lost = 4 (100% loss),

When Jumbo Frames was enabled

  c:\>ping -l 8500 -f 192.168.14.2
  Pinging 192.168.14.2 with 8500 bytes of data:
  Reply from 192.168.14.2: bytes=8500 time<1ms TTL=64
  Reply from 192.168.14.2: bytes=8500 time<1ms TTL=64
  Reply from 192.168.14.2: bytes=8500 time<1ms TTL=64
  Reply from 192.168.14.2: bytes=8500 time<1ms TTL=64
  Ping statistics for 192.168.14.2:
      Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
  Approximate round trip times in milli-seconds:
      Minimum = 0ms, Maximum = 0ms, Average = 0ms

TTL stands for Time-To-Live. Every times packet travels through a node, TTL decreases by one. Hence if TTL is different from each ping request, then packet is taking different routes.