Folks,

 I saw some confusion in this thread. Let's clarify.

 First off, a multi-access (broadcast or NBMA) network N1
 with attached routers R1, R2,...Rz, where Rx is the DR
 is topologically abstracted as a network-LSA, originated
 by Rx, and including links to all R1...Rz, including Rx,
 and type-2 links in the router LSAs originated by R1...Rz.

 As far as the SPT calculation goes, Dijkstra _does_ see
 the network-LSA as a vertex.

 The next-hop calculation for a remote (non-directly connected
 SPT-wise) vertex that is not connected to the calculating
 router (the SPT root) via a transit network is indeed as
 simple as inheriting next-hops from the parent nodes.

 The second case can be explained as follows.

 Physical topology:

     [Ra]
      |
     [Rb] [Rc]
      | |
   ---+------
      |
     [Rd]

  Rd is the calculating router and it does not really matter
  which router is the DR. The abstracted topology representation
  will be as follows:

      +---+
      |Ra |
      +---+
       ^|
       |v
      +---+<--+---+
      |Net|-->|Rc |
      +---+ +---+
       ^|
       |v
      +---+
      |Rd |
      +---+

  So, when Rd calculates the next-hops to Ra, it can see that
  Ra's parent is transit node, so what it does it examines the link
  entries in Ra's router-LSA, finds those referring the netowrk-LSA
  and uses the IP address in Link Data as the next-hop address in
  the route (finding the outgoing interface is easy).

  Regards,
  

-- 
Alex Zinin
Sunday, July 22, 2001, 11:02:51 PM, Sina Mirtorabi wrote:
>>
> Zhang
>>
>> >
>> > N1 is represented by DR, SPF sees only node and p2p
>> > adjacency so what do
>> > you mean by "the parent of R3 is not R2, is N1 "
>>
>> >From the rfc, spf sees only the router-lsa and
>> network-lsa,which router-lsa represents the links of
>> router and network-lsa represents the links of
>> network.
>>
> I'm saying that R1 announce a link to DR ( link ID = DR IP address )
> which in fact represents N1
> if you like, in other words R1 announce a link to the network N1 ( again
> represented by DR ) But my point was if N1 = 192.168.1.0 / 24 with DR (
> 192.168.1.1 ) then the IP subnet is meaning less for SPT and it consider
> a link to 192.168.1.1 which is DR
>>
>> > N1 is just an IP address that does not exist while
>> > we built SPT
>>
>> N1 is represents by the network-lsa,which is generated
>> by the DR.
>>
> see above I meant the IP subnet address, of corse N1 is represented by DR
>>
>> > also R1 does not have a direct adjacency to R3 but
>> > it announce a link type
>> > 2 ( transit ) to the DR and the DR announce a link
>> > to all the attached
>> > routers.
>>
>> DR announce a network lsa, whick includes many links
>> of network, not a link.
>>
>  I meant by " a link to all the attached routers"  >>> it announce a link
> back from DR to every attached router or if you like many link
> ( with many = link to every attached ;-)
> Sina
>>
>> Regards,
>> zhang
>>
>> > so in the directed graph R1 should go through DR in
>> > order to reach R3 so
>> > the parent of R3 is DR ( R2 )
>> >
>> > Sina
>> >
>> >
>> >
>> >
>> > >
>> > >
>> > > Jina.
>>
>> _________________________________________________________
>> Do You Yahoo!? µÇ¼Ãâ·ÑÑÅ»¢µçÓÊ! http://mail.yahoo.com.cn
>>
>> <font color=#6666FF>ÎÞÁÄ£¿ÓôÃÆ£¿¸ßÐË£¿Ã»ÀíÓÉ£¿¶¼À´ÁÄÌì°É£¡</font>¡ª¡ª
>> ÑÅ»¢È«ÐÂÁÄÌìÊÒ! http://cn.chat.yahoo.com/c/roomlist.html

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]