Fragmentation in IS-IS

Unanswered Question
Dec 26th, 2008

Hi,

As we know IS-IS is layer-2 routing protocol, so it doesn't have native support for fragmentation right! So IS-IS use Padding to maximum MTU size to find out the MTU requirement on the link! now again using Padding Router will come to know the connected link MTU! what if we have scenario like: R1---R2---R3, now how R1 will come to know the total path MTU from R1 to R3 if we have different MTU between R1 & R2 as well as R2 & R3? so in short how Fragmentation will work for the IS-IS update packet?

regards

Devang Patel

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (7 ratings)
Loading.
Alejandro Sanchez Sat, 12/27/2008 - 05:27

L2 routing protocol?

I thought it was a common L3 like OSPF, EIGRP, etc...

About fragmentation, don't really know but try to find out something in cisco.

see if the experts would come accross with something...

rgds

Harold Ritter Sat, 12/27/2008 - 09:00

Alex,

I think Devang meant that unlike other IP routing protocols (such as OSPF, BGP, EIGRP, IGRP, RIP, etc), ISIS packets are not carried over IP but rather directly over the Layer 2 protocols.

Regards

Harold Ritter Sat, 12/27/2008 - 08:55

Devang,

There is no concept of LSP fragmentation whatsoever in ISIS. So once the LSP has been originated by a given router, it needs to be able to propagate to all other routers without fragmentation. To achieve that, you either need to have the same MTU everywhere in the network or to set the "lsp-mtu" to the value of the smallest MTU in the network so that all routers agree what is the largest LSP that can be flooded through the network. Note that the default value is 1497.

Refer to the following document for more information on the command "lsp-mtu":

http://www.cisco.com/en/US/docs/ios/isoclns/command/reference/iso_a1.html#wp1021123

Regards

devang_etcom Sat, 12/27/2008 - 10:23

hi Hritter,

Thanks for your reply! so you said all routers have to agree for the smallest MTU in the network! but how those routers will communicated about the smallest MTU size in the network?

what kind of message they will use to discover the MTU of the path!

regards

Devang Patel

Harold Ritter Sun, 12/28/2008 - 07:26

Devang,

There is no automatic way to achieve that. This is part of network planning.

Regards

devang_etcom Sun, 12/28/2008 - 11:30

"This is part of network planning. " so does it mean during the network planning and design engineers has to take care of it?

regards

Devang Patel

Giuseppe Larosa Sun, 12/28/2008 - 12:52

Hello Harold,

an IS-IS LSP can be "fragmented" but only by the LSP owner/originator.

the LSP PDU format includes a field called LSP ID.

An LSP-ID is made of the following fields

system-id. 1 byte that says if originator is a node or a pseudonode (0x00 if router) a byte called LSP number and a 4byte Sequence number.

the byte called LSP number allows to carry information of a single LSP within multiple IS-IS PDUs each with incremental LSP number field.

so there is no critical MTU problem for LSP.

At least in theory.

However, an MTU mismatch is a problem because hello PDUs are padded to MTU value.

I see that the command lsp-mtu allows to reduce the LSP MTU and is required if there is a link with MTU less then 1500 bytes.

For the link state nature of IS-IS the command if needed has to be applied to all routers not only the two on the link with lower MTU

I understand now what you mean: once the LSP has leaved the originator node it cannot be fragmented by any other node : or it is able to travel over the network topology (LSP flooding) or it will be lost counting as a giant.

Hope to help

Giuseppe

devang_etcom Sun, 12/28/2008 - 13:32

Hi Giuseppe,

so again if there is a link that supports smaller MTU then 1500bytes then I will have to configure lsp-mtu on all routers in the network! as only originator can do the fragmentation then LSP will never reach to the destination! as it will be counted as a giant and it will be in counter of sh interface as a input error!!! is that correct?

regards

Devang Patel

Harold Ritter Sun, 12/28/2008 - 18:36

Devang,

That is correct. If a link with a MTU smaller than 1500 existed in the network, you would need to change the default lsp-mtu value throughout the network. It is very unlikely that this condition would exist on recent networks though.

Regards

Harold Ritter Sun, 12/28/2008 - 18:08

Giuseppe,

That is precisely what I meant. Once the LSP (more than one fragment if one is not enough) is originated, it can not be fragmented. It will not be an issue in most network though as the default value for the lsp-mtu is 1497 and the fact that most networks will not have links with smaller MTUs.

The issue with mismatch MTU is different and will be avoided by the hello padding as you mentioned. The adjacency will simply not come up if the MTU size is not the same.

Regards

Actions

This Discussion