OSPF on point to point networks

Unanswered Question
Jun 8th, 2007
User Badges:

Hi,

Could someone clarify this point...


The destination address of OSPF packets on Point to Point networks will always be 224.0.0.5(a multicast address).


Why a router has to multicast an OSPF packet rather than to unicast it as there will be only one destination on this network?


Thanks in advance,

Pavan

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 3.5 (2 ratings)
Loading.
mohammedmahmoud Fri, 06/08/2007 - 03:54
User Badges:
  • Green, 3000 points or more

Hi Pavan,


Since the neighbor is not manually configured with a neighbor command in this scenario, the OSPF process won't know the unicast address and thus it will use multicast (instead of broadcast which was used in RIPv.1 for example).



HTH, please do rate all helpful replies,

Mohammed Mahmoud.

pavansuraj Fri, 06/08/2007 - 04:10
User Badges:

Hi Mohammed,

Thanks for your response.

As the neighbor is directly connected and probably in the same subnet which it's own interface is located in, Why can't the router assume its address?

like, the router sending a packect to a destination in both cases having either an exit interface or a next-hop address in its routing table.

Thanks once again

Pavan


Harold Ritter Fri, 06/08/2007 - 04:23
User Badges:
  • Cisco Employee,

Pavan,


The router doesn't need to assume or guess what the neighbor address is. It simply needs to use the multicast address as the destination address, push the packet through the serial line and sure enough the packet will make it to the neighbor, which will know for sure the packet is destined for it.


Why should the router guess the neighbor IP address. What if a /24 is used on that link? An worse what if ip unnumbered is used on that link? Using the multicast address is a simple way to alleviate all that.


Hope this helps,

mohammedmahmoud Fri, 06/08/2007 - 04:26
User Badges:
  • Green, 3000 points or more

Hi Pavan,


I guess that the answer to your question is divided in to 2 parts, the first is, this is how the algorithm is written and the second is because point-to-point link doesn't always obey this rule, as some times for example we use unnumbered interfaces and thus using multicast would be the generalized solution.


http://www.cisco.com/en/US/tech/tk365/technologies_configuration_example09186a0080094057.shtml



HTH, please do rate all helpful replies,

Mohammed Mahmoud.

pavansuraj Fri, 06/08/2007 - 21:27
User Badges:

Hi Mohammed and Harold,


Thankyou very much for your input.


Best Regards

Pavan

purohit_810 Fri, 06/08/2007 - 05:48
User Badges:
  • Silver, 250 points or more

Hi Pavan,


1) Unicast Addresses are Not reserved address.. People might can use the same addresses into the Network.


2) Application developer (Protocol Developer) has to define IP address for thye same activity. You cannot modify it. (Are able to Modify 224.0.0.5 by another else IP address).


3) Multicast addresses are mint(Reserved for this perpuse only) for that.]


4) Unicast IP address is always changes in charatristics mean


192.168.10.160/24 (This unicast IP if you are using /24 subnetmask)


192.168.10.160/27 (This is Network ID for /27).


Regards,

Dharmesh Purohit

Actions

This Discussion