> oh is it. i thought virtual links is a good idea.

Virtual links is just a hack to overcome some most obvious consequences
of limiting area hierarchy to be 1 level deep (backbone as root and
everything else is leaf). In the long term hacks rarely make any good.
I agree with Acee, on new design it would have been better to place
complexity somewhere else. For example, in allowing true area hierarchy.

Anton


On 11/16/2010 03:42 AM, p6 c6d6 wrote:
> 
> On Mon, Nov 15, 2010 at 3:20 PM, Acee Lindem <[email protected]
> <mailto:[email protected]>> wrote:
> 
> 
>     This is due to the fact that paths via transit area ABRs other than
>     the virtual link endpoints can be installed. If I were designing a
>     link-state protocol from scratch, I would not include the complexity
>     of virtual links as they are specified in OSPF.
> 
>     Hope this helps,
>     Acee
> 
>     On Nov 14, 2010, at 9:47 PM, p6 c6d6 wrote:
> 
>     > hi,
>     >
>     > why should one recalculate routes using transit areas summary
>     lsas, even if
>     > better paths exist through transit area.  as a virtual link is
>     considered as
>     > an unumbered p2p link, it is as good as the router has an
>     interface to the backbone.
>     > referring to rfc2328 figure 17, RT1 chooses route network N1
>     through RT4, since that
>     > is an intra-area path, that is in accordance to the path
>     preference rules of ospf.
>     > why do we give special treatment for transit areas. If the link
>     were not a virtual-link and
>     > if it indeed is real p2p interface in to the backbone, we will not
>     try to route the data to N1
>     > through RT5, since the route through RT4 would be preferred. why
>     special treatment for
>     > transit areas, and why can't we extend this treatment to other
>     areas as well (neglecting
>     > the fact that area routing is protected by bad routing of other
>     areas).
>     >
>     > thanks in advance
>     >
>     >
>     > <ATT00001..txt>
> 
> 
> oh is it. i thought virtual links is a good idea.
> and for the newly calculated route  why do we
> have to use the associated area as backbone area
> though the path is still through transit area ?
> 
> 
> 
> 
> _______________________________________________
> OSPF mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/ospf
_______________________________________________
OSPF mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ospf

Reply via email to