Alex-

Yes, in your 2nd scenario you can get transient loops
in the inter-area routing. See Section 7 of Alex (Zinin)'s
expired shortcut ABR draft, which I have posted at
ftp://ospf.org/specs/ospf-shortcut-abr-02.txt for more details.

As for your third scenario, I personally suppress the summaries
only if *all* the next hops belong to the area. So in this case, I would
generate summaries for both Areas 1 and 2 (never for the backbone though,
as the route to X is associated with the backbone). For unicast, suppressing
the summaries is simply an efficiency thing. And for MOSPF, I think
either strategy (suppress if *all*, or suppress if *any*) works.

John

> -----Original Message-----
> From: alex lerin [mailto:alexlerinYAHOO.COM]
> Sent: Tuesday, January 22, 2002 12:07 AM
> To: OSPFDISCUSS.MICROSOFT.COM
> Subject: Re: Origination of summary LSA into the transit area
>
>
> Hi,
> I need more clarification for the 2nd & 3rd scenarios
>
> In the second scenario, if the intra area cost to X
> increases in ABR2, the intra-area cost would get
> over-written by the lesser cost in the transit area
> summary lsa(If destination X is directly connected to
> ABR2, ABR1 will not get the new summary lsa with
> increased cost till the SPF calculation is complete).
> This will go on in a loop till the summary lsa
> originated by ABR2 in transit area reaches the actual
> cost configured to the directly connected destination
> X.
>
> For the thrid scenario, RFC tells that if the next hop
> of the equal cost paths lie in the area, do not
> originate summary LSA into that area. If suppose from
> ABR1 there are equal cost paths to destination X
> through Area A1, Area A2 and backbone area(with next
> hop in Area A1, Area A2 & backbone area), how are we
> to interpret step 4 in section 12.4.3
>
> rgds,
> Alex
>
>
> --- Alex Zinin <azininNEXSI.COM> wrote:
> > alex,
> >
> > > I have a query in section 12.4.3(Origination of
> > > summary LSAs). Step 4 in secion 12.4.3 indicates
> > that
> > > if the next hop of the route falls in the area
> > > A(transit capable), the summary LSA for that
> > > destination must not be originated in area A.
> >
> > > Consider the first scenario. ABR1 is connected to
> > Area
> > > A1 & Area A2. ABR2 is connected to Area 1 , Area 2
> > &
> > > Area 0.
> > > ABR1 & ABR2 are virtual end-points. Area 1 & Area
> > 2
> > > are the two transit areas. The cost from ABR1 to
> > ABR2
> > > in Area 1 & Area 2 are equal. So the cost of the
> > > virtual link are equal.
> >
> > > ABR2 has an intra- area route to a destination X.
> > > ABR1 will also have 2 equal paths to destination
> > X(one
> > > through Area 1 & Area 2.
> > > Should ABR1 originate summary LSA for destination
> > X in
> > > Area 1 & Area 2? This would go against the split
> > > horizon mechanism.
> >
> > No, it should not originate them.
> > ABR2 will do so though, so both internal routers
> > and ABR1 will use its summaries to calculate/update
> > the routes to X.
> >
> > > Now suppose there is a second scenario where the
> > cost
> > > from ABR1 to ABR2 is not equal. The cost through
> > Area
> > > A1 is lesser than the cost through Area A2. Now
> > ABR1
> > > will have only one path to destination X through
> > area
> > > A1. so ABR will originate summary LSA for
> > destination
> > > X in area A2. since Area A2 is also transit
> > capable,
> > > ABR2 will consider summary LSA originated by ABR1
> > in
> > > Area A2. This would lead to path loop to
> > destination
> > > X.
> > > What is the method to resolve this?
> >
> > ABR1's summary will not update ABR2's route to X
> > because of the cost.
> >
> > > Also if ABR1 is connected to backbone area
> > through
> > > another interface, it would have a route to
> > > destination X with next hop in backbone area
> > itself.
> > > Now should ABR1 originate summary LSA for
> > destination
> > > X in Area A1 & Area A2. Will this summary LSA be
> > used
> > > by any of the normal routers in Area A1 & Area A2.
> >
> > Yes, it should originate the summaries.
> > Wether they will be used or not will depend on
> > the results of comparison with the summaries
> > originated
> > by the other ABR.
> >
> > Alex.
>
>
> __________________________________________________
> Do You Yahoo!?
> Send FREE video emails in Yahoo! Mail!
> http://promo.yahoo.com/videomail/
>

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