10-31-2013 11:10 AM - edited 03-07-2019 04:21 PM
Hi All,
The show interface fax/x output contains a feild called "giants" which according to cisco means "Number of packets that are discarded because they exceed the medium's maximum packet size" My question is the packet size exceeds the maximum packet size will it not be fragmented ? Why does the interface have to drop it, or does this counter only contain packets exceding the size and have the DF bit set ??
Thanks in Advance
Umesh Shett
10-31-2013 12:14 PM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
Fragmentation happens on L3 egress (unless packets sets the DF bit).
L2 assumes frames, both ingress or egress, don't exceed MTU.
11-03-2013 04:19 AM
Thnx Joseph,
What if a router receives a 1700 Byte frame ? Wikepedia says that the maximum payload size of an ethernet is 1500 octets (which is also the egress interfaces MTU) if the routers ingress ethernet interafce starts dropping frames that exceed the 1500 + 18 (ethernet encapsulation) limit there would be no frangemntation ever needed. At no times the ingress interface will ever send an L3 packet which is more that 1500 octets and hence will never exceed the egress interfaces MTU and hence no fragmentation. This is what I am confused about , hope you can help.
Regards
Umesh
11-03-2013 05:34 AM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
If the receiving (ingress) port was "standard" Ethernet, the received frame would be dropped, as being oversized.
If the receiving (ingress) port was "jumbo" capable Ethernet, jumbos were enabled, and if its jumbos supported 1700 bytes (there's no jumbo size standard), then the frame would be accepted on the port. What happens next would depend whether the received frame needed to be forwarded (it might be directed to the router itself), and if so, how the egress port was configured. It might be dropped by an egress port on that router, forwarded as same size, or L3 fragmented into multiple L2 frames.
BTW, fragmentation is at L3, i.e. packets are fragmented, Ethernet frames are L2 and are never fragmented. Even from the source host, with standard Ethernet end-to-end, packets will be fragmented if they are larger than the frame can contain. Normally, source host will size its packets to fit into the host's frames without the need to fragment.
11-03-2013 06:08 AM
Hi Joesh,
Just assuming a network shown below if I were to ping from host A to Host B with 1600 bytes of data. That would be a total of 1600 (data) + 8 bytes (ICMP header) + 20 bytes ( IP Header) + 18 Bytes (Ethernet Header) == 1646 Bytes of ethernet frame. So when R1's ingress interface receives the frame and decapsulates it , it must have an L3 packet of 1628 bytes and If I do not set the df bit in the ICMP packet i would expect the packet to be fragmented on R1's egress interfaces(connected to R2).
Thanks in Advance
Umesh
11-03-2013 04:27 PM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
For what you described to work as described, would require host 1 (in you drawing or host A in your text) and R1's interface to that host to support jumbo Ethernet and the rest of the interfaces standard Ethernet.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide