09-01-2005 09:59 AM
Hello,
I am deploying a EoMPLS service with a service provider in his proudction network.
The Core routers on this network are configured with the "mpls mtu 1508". This MPLS MTU size does not work with an EoMPLS implementation, I need at least 1530 bytes, so I need to change the MPLS MTU command on all the Core Routers.
My question is: Is there any impact of configuring the "mpls mtu 1530"? Does this affect the LDP sessions I have? I think not but I have read the following:
"All devices on a physical medium must have the same MPLS MTU value in order for MPLS to interoperate"
Obviously, I will configure the "mpls mtu" on one router first and then on the neighbors, but one at a time (it is an ethernet segment) so ... will the LDP session remain up and running for the time it takes to configure all the routers?
Thanks for your responses!!
Alex
09-01-2005 10:23 AM
Configuring "mpls mtu 1530" has no impact. You just need to make sure that the layer 2 devices connecting the routers support ethernet frames of 1530 bytes.
The LDP session will not be affected while you are migrating from 1508 to 1530.
Let me know if I answered your question,
09-02-2005 01:51 PM
Hi,
Sorry to barge in,but I have three questions:
1. If you do not configure 'mpls mtu' command on the interface is it true that default cisco behavior is to set mpls mtu to be ip mtu + 8 bytes?
2. when router receives a packet (IP or non IP) which is larger than mtu configured in ingress interface, the packet is dropped (as opposed to fragmenting packet, if allowed, on egress interface).
Is it true that if mpls is enabled on the interface router will fragment the packet on ingress interface as well (based on mpls mtu) and not drop it?
3. when passing labeled packets (1508 bytes and up) over switches interconnecting router, what do you have to enable on switches not to drop the packet as being "baby giants"? do you have to enable jumbo frames support to make it work or maybe something else?
Thanks,
David
09-02-2005 02:35 PM
1. the default bahavior is that the mpls mtu is the same as interface mtu (default of 1500 for Ethernet interfaces).
2. If "mpls mtu" is set to a higher value than 1500 on the igress interface then the incoming baby giant frames will be accepted if they contain an MPLS information. No fragmentation is required at this point.
3. This is different on just about every switches. Here's a good URL to help you determine what you need to configure on different Cisco switches:
Let me know if I answered your questions,
09-03-2005 02:01 AM
Hi,
Thank you for your answer. Let me reiterate regarding point number 2 above.
My question is if mpls mtu is set for 1500 on the ingress interface and double tagged frame arrives (1508 bytes), will a router fragment a frame to two double tagged tagged frames or will it drop it until mpls mtu of 1508 and above is configured on this ingress interface?
Thanks,
David
09-03-2005 05:19 AM
In this case, the frame would be dropped. Fragmentation is typically performed on the outbound interface, not on the inbound.
As you indicated, setting the "mpls mtu 1508" on the ingress interface will solve the issue.
Let me know if I answered your question,
09-03-2005 06:53 AM
Just to complete on my explanation, if the outbound interface on the ingress router had a mpls mtu of 1500 then fragmentation would take place assuming that an IPv4 packet was received and that the DF bit is not set.
Hope this helps,
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: