Jump to content


OSPF, distance vector ?


  • Please log in to reply
11 replies to this topic

#1 pappyaar

pappyaar

    Cisco Routing/EEM/TCL

  • Technical Experts
  • PipPipPipPip
  • 959 posts
  • Gender:Male

Posted 19 September 2008 - 05:27 AM

Hi all, i know all of the CCNP are aware of the fact that besides being pure classless routing protocol, OSPF does have some distance vector properties, rather it would not be wrong to say that OSPF does react like a completely distance vector routing protocol in a certain situation !

What situation might that be, and how would you prove that on IOS ?

Those of you who dont have any idea of it but still want to answer it, can surely pm me to get some hints :-)
  • 0

#2 karanc

karanc

    Newbie

  • Members
  • Pip
  • 28 posts

Posted 19 September 2008 - 07:40 AM

i think you have to play with the metric of ospf to make all the links of same cost .....then it will behave like distance vector ......just like if we leave the default reference bandwidth as 100 Mbps ...then thr will be no difference in 1 Gbps and 10 Gbps or 100 Gps....

so if i change the reference bandwidth according to my structure needs to make all link speed correspond to one ...i can make ospf somewhat look like distance vector ....

well rest we can hear from our gr8 friends here ......
  • 0

#3 faisal.saleem

faisal.saleem

    $ïιεη† ⊕pεrα†⊕r

  • Veterans
  • PipPipPipPipPip
  • 1244 posts
  • Gender:Male
  • Location:Kuwait
  • Interests:IT/Telcom

Posted 19 September 2008 - 08:46 AM

Hmm that is another interesting question :) ok let me try, OSPF is a distance vector protocol when it deals with in 1 area and An ospf router in an area can not see the other prefixes beyond the Area Border Router so it depends on the Area Border Router and we can say that like in RIP etc we always depends on the next router what he tell we assume he is right. So if some 1 will ask me this question that OSPF is distance vector or link state then i'll always answer that if the scenario is dealing only with inter area OSPF routing then the OSPF is a distance vector routing protocol...

Well i know my answer is not correct 100% what the pappyaar originally demanded but i just tried to answer it because it is also another view of this concept..

Faisal.
  • 0

#4 talent pk

talent pk

    Cisco Newbie

  • Technical Experts
  • PipPipPipPipPip
  • 4026 posts
  • Gender:Male
  • Location:Pakistan
  • Interests:Oracle

Posted 19 September 2008 - 08:38 PM

well faisal all routers inside their own area knows all the possible paths, so they are in a linkstate,

about OSPF being DistanceVector: Assuming that we have Redundant Routers between 2 Areas or ABRS with redundant connections. Is this situation happens sometimes when we have a route flapping issues in one area causing SPF to run in that area while getting the same route as Summary LSA back to its area (which assumed to be router LSA instead of summary LSA) and then the route is wrongly assumed to be coming from a source which is not what it is. i.e. the other Area ?
i have heard from someone before. but i dont remember.

however i think this is Routing by Rumour behaviour. rather a Distance Vector.

Edited by talent pk, 19 September 2008 - 08:47 PM.

  • 0

#5 Lord Flasheart

Lord Flasheart

    All Rise For The Lord

  • Veterans
  • PipPipPipPipPip
  • 2848 posts
  • Gender:Male
  • Location:England

Posted 19 September 2008 - 09:44 PM

Inter Area OSPF behaves like distance vector. The ABR says to routers in one of it's areas "Hey, I know how to get to a.b.c.d etc" using Type 3 LSAs. This is distance vector behaviour. As we know, distance vector behaviour can lead to routing loops hence all traffic must pass through Area 0 (loop-free).
  • 0

#6 pappyaar

pappyaar

    Cisco Routing/EEM/TCL

  • Technical Experts
  • PipPipPipPip
  • 959 posts
  • Gender:Male

Posted 20 September 2008 - 12:50 AM

well faisal all routers inside their own area knows all the possible paths, so they are in a linkstate,

about OSPF being DistanceVector: Assuming that we have Redundant Routers between 2 Areas or ABRS with redundant connections. Is this situation happens sometimes when we have a route flapping issues in one area causing SPF to run in that area while getting the same route as Summary LSA back to its area (which assumed to be router LSA instead of summary LSA) and then the route is wrongly assumed to be coming from a source which is not what it is. i.e. the other Area ?
i have heard from someone before. but i dont remember.

however i think this is Routing by Rumour behaviour. rather a Distance Vector.


Very brilliant answer, thanks for the feedback all of you :-) this post will be long, becoz i observed something very intersting and would like to share with all of you

consider this diagram

R1-----------------------R2-------------------------R3

consider a distance vector ( i will call it DV from now on ) protocol like EIGRP. lets say R1 advertises 2 routes 100.0.0.0/8 and 101.0.0.0/8 to R2. now by default R2 will advertise both these routes to R3 right ?? now if i filter the route 100.0.0.0/8 on R2 will R2 still advertise this route to R3 ??? answer is NO, DV protocols consider strictly whats in their routing table and only advertise what they know, OSPF ( within area only ) consider whats in their LSDB rather then their routing table.
If i run OSPF in the above diagram, and filter 100.0.0.0/8 at R2, R2 will STILL advertise this prefix to R3, why ?? becoz OSPF link state algorithm says that the LSDB of every router within an area MUST be same, when the routes are propagated from some other area, ABR of the receiving area treats them just like DV routes. how ?? to prove it consider the diagram

R1-------------------------------------R2
|
|
|
|
|
|
|
R3------------------------------------R4

R1 and R2 are in area 1,
R1 and R3 are in area 0,
R3 and R4 are in area 2,

routes coming from R2 will be seen as O IA on R3 right ??? now if you will filter these routes ( or replace the OSPF learned routes via a static route, so that its no longer as a OSPF route in routing table) the routes from R2 will no longer be propagated to area 2 !!!!!!

Now i will invite you to go through OSPF RFC, it says that an ABR will originate type 3 LSA looking at the routing table only :-) so there is not much LS about type 3 LSA at all :-). Most of your confusion will be cleared if you read the RFC under the originating type 3 LSA portion, its a must read.

I hope this clears :-)
  • 0

#7 faisal.saleem

faisal.saleem

    $ïιεη† ⊕pεrα†⊕r

  • Veterans
  • PipPipPipPipPip
  • 1244 posts
  • Gender:Male
  • Location:Kuwait
  • Interests:IT/Telcom

Posted 20 September 2008 - 12:52 AM

well faisal all routers inside their own area knows all the possible paths, so they are in a linkstate,


The behavior of forming adjacency to other routers you can say it is Link State but the operation as i and flash heart wrote is definitely a distance vector behavior, basically i was also not aware of that fact because it is not written in the books but once upon a time i read this on networkworld in a post of my fav author ie jeff doyle, and there was good debate on this point. Jeff is a regular poster on this community and you can find enough good info in his articles ==> www.networkworld.com/community/doyle

Now tried to find but forgot about his original article on this issue.
  • 0

#8 talent pk

talent pk

    Cisco Newbie

  • Technical Experts
  • PipPipPipPipPip
  • 4026 posts
  • Gender:Male
  • Location:Pakistan
  • Interests:Oracle

Posted 20 September 2008 - 05:42 AM

yes faisal i agree with your points, it was your first line that made me confuse, becoz inside one Area all the routers share Router LSA or LSA-Type1 that must agree on same LSDB synchronization, all routers knows about all the open paths to reach eachother.
  • 0

#9 talent pk

talent pk

    Cisco Newbie

  • Technical Experts
  • PipPipPipPipPip
  • 4026 posts
  • Gender:Male
  • Location:Pakistan
  • Interests:Oracle

Posted 20 September 2008 - 05:44 AM

Special Thanks to Pappyar:
==================

Dear Friend, i find your way of querying more efficient way of learning then others. Please keep going and faisal and i really appreciate your knowledge Mashallah.

thanks once again
  • 0

#10 rainbow9810

rainbow9810

    Cisco Expert

  • Veterans
  • PipPipPipPipPip
  • 1865 posts
  • Gender:Male
  • Location:Gods own country

Posted 02 October 2008 - 08:16 PM

Congrats Pappyaar for becoming "Best Poster -September 2008" .. All the best....
  • 0

#11 popeye77

popeye77

    Newbie

  • Members
  • Pip
  • 6 posts

Posted 17 June 2009 - 05:34 AM

Yes, Pappiar got it right. To elaborate more on that.

You have area 1, 0 and area 2. If you have a link failure in area 1 that will not trigger a SPF run on other areas. The type 3 lsa which was originated from area 1 based on a link( which is failed now) will be flooded to other area with infinite metric. The other routers in area 0 and area 2 will add cost 1 to the max and becomes infinite metric.

This is called partial run of SPF. The SPF is run on the local area (affected area) and uses distance vector to flood the route with infinite metric(Type 3LSA routes) to other areas.

Hope this helps.




  • 0

#12 popeye77

popeye77

    Newbie

  • Members
  • Pip
  • 6 posts

Posted 17 June 2009 - 05:40 AM

Yes, Pappiar got it right. To elaborate more on that.

You have area 1, 0 and area 2. If you have a link failure in area 1 that will not trigger a SPF run on other areas. The type 3 lsa which was originated from area 1 based on a link( which is failed now) will be flooded to other area with infinite metric. The other routers in area 0 and area 2 will add cost 1 to the max and becomes infinite metric.

This is called partial run of SPF. The SPF is run on the local area (affected area) and uses distance vector to flood the route with infinite metric(Type 3LSA routes) to other areas.

Hope this helps.




  • 0





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users