System MTU explained

By default any interface on the switch can only send or receive a frame of a maximum size 1500 byte. Some applications like (QinQ, MPLS over Ethernet) require increasing the MTU of the switches in your network to allow greater sizes because thier frames may exceed the default 1500 bytes size.

You can increase the MTU of all interfaces operating at 10/100 Mbits using the command system mtu <vlaue>. This command will not affect GigaEthernet ports; the command system mtu jumbo may be used to alter the MTU of the GigaEthernet interfaces.

You will need to reload the switch for the new value to take effect.

finally you can change the MTU of routed ports using the command system mtu routing <value> and it can not be set to a value greater than the system MTU.

Verification Example:

Swtich1 is directly connected to Switch2 using interface f0/13 on both switches. Switch1 system MTU was set to 1600 and Switch2 is set to 1550 as shown in the configuration below:

!-- switch1 Configuration showing the MTU
SW1#sh ip int b | in Vlan
Vlan1                  192.168.12.1    YES NVRAM  up                    up

SW1#show system mtu

System MTU size is 1600 bytes
System Jumbo MTU size is 1600 bytes
Routing MTU size is 1600 bytes

!-- switch1 Configuration showing the MTU

SW2#sh ip int b | in V
Vlan1                  192.168.12.2    YES NVRAM  up                    up

SW2#show system mtu 

System MTU size is 1550 bytes
System Jumbo MTU size is 1550 bytes
Routing MTU size is 1550 bytes

SW2#sh int f0/13 | in (MTU|giants|packets )
  MTU 1550 bytes, BW 100000 Kbit, DLY 100 usec,
     31 packets input, 3163 bytes, 0 no buffer
     0 runts, 0 giants, 0 throttles
     0 input packets with dribble condition detected
     316 packets output, 23747 bytes, 0 underruns

Operation verification:

!-- Ping from SW1 with size 1550 succeeded

SW1#ping 192.168.12.2 size 1550 df-bit 

Type escape sequence to abort.
Sending 5, 1550-byte ICMP Echos to 192.168.12.2, timeout is 2 seconds:
Packet sent with the DF bit set
!!!!!
Success rate is 100 percent (5/5), round-trip min/avg/max = 1/202/1006 ms

!-- Ping from SW1 with size 1600 failed on SW2 as shown below

SW1#ping 192.168.12.2 size 1600 df-bit

Type escape sequence to abort.
Sending 5, 1600-byte ICMP Echos to 192.168.12.2, timeout is 2 seconds:
Packet sent with the DF bit set
.....
Success rate is 0 percent (0/5)

!-- interface f0/13 on Sw2 shows giants received = the number of failed ping packets

SW2#sh int f0/13 | in (MTU|giants|packets )
  MTU 1550 bytes, BW 100000 Kbit, DLY 100 usec,
     82 packets input, 23609 bytes, 0 no buffer
     0 runts, 5 giants, 0 throttles
     0 input packets with dribble condition detected
     749 packets output, 64729 bytes, 0 underruns

To recap, pinging Sw2 with a packet size greater that its configured system mtu of 1550 has passed from Sw1 because its "MTU is 1600" but was not accepted by Sw2 and considered as Giant frames.