cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1720
Views
0
Helpful
6
Replies

MSS and MTU

jonk34567
Level 4
Level 4

 

Hi team,

IN our network we have to change the mtu size due to GRE overhead ( 12 bytes ). For that i have come with 2 commands  Do i need to enter both commands or one if enough ? 

 

Current IP MTU(1500) = IP Header ( 20 ) + TCP header ( 20 ) + MSS ( 1460 )

New IP MTU (1488) = IP Header ( 20 ) + TCP header ( 20 ) + MSS ( 1448 )

Ethernet MTU ( 1514 ) =Ethernet Header (14) + GRE Header ( 12 ) +  IP Header ( 20 ) + TCP header ( 20 ) + MSS ( 1448 )

 

Option-1 

Change IP MTU to 1488

Router1(config)# int ethernet 1/0
Router1(config-if)# ip mtu 1488

 

Option-2

Change MSS to 1448

Router1(config)# int ethernet 1/0

Router1(config-if)# ip tcp adjust-mss 1448

 

Do we need to go with both options or either one of them is enough ? 

 

 

6 Replies 6

Giuseppe Larosa
Hall of Fame
Hall of Fame

Hello @jonk34567 ,

both options because not all traffic is based on TCP for example a video stream would use UDP.

 

Hope to help

Giuseppe

 

Joseph W. Doherty
Hall of Fame
Hall of Fame

BTW, GRE adds 24 bytes, as the GRE tag, itself is 4 byte, but another IP header (20 bytes) is added.

Also BTW, IP TCP adjust-mss only needs to actually be placed on one router, it deals with traffic both in and out (although I too generally place in on both routers).

You might also consider enabling PMTUD on the router, to deal with a link between the GRE end points that doesn't support 1500 MTU. (Effectively, the GRE packets get DF set.)

Also keep in mind, the 20 byte sizes for IP header and TCP headers are minimums. Either or both can be larger (another reason for PMTUD).

A Cisco paper on dealing with encapsulation fragmentation is: https://www.cisco.com/c/en/us/support/docs/ip/generic-routing-encapsulation-gre/25885-pmtud-ipfrag.html.

Hi,

 

If two routers are communicating with diffrent MTU sizes what will happen ? will they choose the lowest amoung them or they will communicate with their own vaules.

Two devices on the same physical link should have the same MTU, however if they don't (and assuming the media can accept the larger of the two device's MTU), the smaller can send to the larger, no problem. The larger sending to the smaller may raise an oversize frame on the smaller (which may consider the frame corrupted - and drop it).

If the same device receives an ingress frame, on one interface, larger than the MTU of a L3 egress interface, the interface will fragment the packet, if DF not set, or drop it if DF is set. For the latter, it will send an IGMP message back to the sender that the packet needed to be fragmented. (On old implementations, that's all the IGMP message indicated, the newer implementations also send the max MTU supported. [The latter is a great improvement as the former required the sender to try different sizes until it found a size that could be transmitted.])

Hi Joseph,

 

Are suggesting to configure PMTUD with below commands. Also if i execute below command does it affect only tcp traffic and ICMP traffic still be frangmented ? 

ip tcp path-mtu-discovery

Yes, that's the command. It only impacts traffic sourced from the router (which will be the case for a tunnel endpoint defined on the router). It impacts all IP traffic (again, sourced from the router).
Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco