Basic IS-IS questions

CriscoSystems · ‎10-18-2009

I took the "Do I know this already" quiz in Cisco's BSCI Exam Certification Guide, and one of the questions asked me to identify which statements were true about IS-IS. One of them was that all L2 routers must be contiguous, which I selected as true, and the answer key said it was wrong. Later on in the chapter it is stated that in fact L2 routers DO need to be contiguous (makes sense to me).

So don't they? WHY WHY WHY WHY WHY does Cisco Press publish exams to make people who are actually right think they're wrong?!?!

I also encountered the claim that all L1 routers within an area form adjacencies with each other; while only the DIS sends out LSP's. (Unlike in OSPF where they only get adjacent with the DR and BDR.) Well then why do non-DIS routers form adjacencies with each other? What exactly does adjacency mean in the IS-IS context?

Giuseppe Larosa · ‎10-18-2009

Hello Stuey,

Cisco books are written by human beings so they can contain errors even in the "Do I know this already" section.

Cisco implementation strictly requires a L2 backbone formed by contiguous L2 links.

IS-IS standard has a feature for a partitioned backbone that has a name like auto-repair or auto-restore but this is not supported in Cisco implementation of IS-IS.

So if the question would be

" does IS-IS standard require that all L2 nodes are contiguous?"

the answer is no

if the question would be:

" Does Cisco IS-IS implementation require that L2 nodes are contiguous?"

the answer is yes

About DIS:

all devices become adjacent with pseudo-node DIS.

the DIS device builds an LSP for the psuedonode where all attached nodes appear listed.

DIS device sends its hellos every 3.3 seconds and a short listing of database with all LSP headers every 10 seconds.

Summing all these facts I've the impression that IS-IS DIS works more then OSPF DR.

the meaning of adjacency is the same as in OSPF: link state DB is synchronized.

Hope to help

Giuseppe

Peter Paluch · ‎10-18-2009

Hello Giuseppe,

Very nice reply indeed. Thanks for that. Please allow me to have a couple of comments.

I am afraid that even the IS-IS standard (the ISO/IEC 10589:2002) does not allow or provision for the Level-2 partition repair. The IS-IS standard talks only about Level-1 partition repair through Level-2 domain. If a Level-2 domain (the backbone) becomes partitioned, there are no provisions for its healing in IS-IS itself and the connectivity will be impaired. The IS-IS standard can be downloaded from this page free of charge:

http://standards.iso.org/ittf/PubliclyAvailableStandards/index.html

My second thought goes to whether the DIS has more work than DR in OSPF. At first sight, it has - it needs to send hellos three times faster than remaining routers on a segment, and it transmits the "table of contents" of its link-state database every 10 seconds, exactly as you have suggested. However, that's about it. There is almost nothing more than just this. The DR in OSPF does not transmit the "table of contents" periodically but whenever a new router comes on a link, it has to go through a fairly complicated synchronization procedure. Actually, the need of a DR to maintain a state machine for each neighbor and the requirements of the OSPF's transport protocol during database synchronization account for the significant part of the OSPF's complexity. So in terms of periodic and routine work, yes, the DIS has slightly more work to do. However, regarding the other responsibilities of being DIS/DR, it appears to me that being a DR is more complicated than being a DIS.

Best regards,

Peter

Giuseppe Larosa · ‎10-19-2009

Hello Peter,

thanks for your correction I didn't check documentation before writing my post so I thought auto-repair could apply to L2.

rated as it deserves

Hope to help

Giuseppe