Re: NSSA Problem

[Acee Lindem <[e-mail]>]

Amit Srivastava wrote:

> Hi Acee,
>       I am sorry but i would not agree to your comment
> See when multiple ABRs are there each of them would
> have given router LSA in both the Area and when it
> gets an AS-External LSA it can simply send the LSA to
> other Area and will also give summary LSA(Type 4) for
> the LSA.


Amit,

But an ABR for an NSSA does not generate type 4 summaries
for ASBRs originating type 7 LSAs within the NSSA. This
is clarified in section 2.3 of draft-ietf-ospf-nssa-update-11.txt.
This design point is consistent with the goal of isolating the
NSSA from the rest of the OSPF routing domain.

Thanks,
ACee



> Now the router who will receive this LSA
> would just do the SPF calculate the shortest ABR and
> forward the packet to that ABR.
>         That is why i am not able to appriciate the
> Must condition of having forwarding address in the
> type 7 LSAs.
> Regards
> Amit
>
> --- Acee Lindem <[e-mail]> wrote:
>
>>Amit Srivastava wrote:
>>
>>
>>>Hi Pat and Acee,
>>>      See even if the forwarding address is not
>>>
>>set in
>>
>>>the originating AS-external LSA(in the NSSA) the
>>>
>>ABR
>>
>>>can know about the router which has originated the
>>>
>>LSA
>>
>>>by its router id. So why they are imphasising on
>>>setting of the forwarding address!!
>>>I am still in the same state of confision plz
>>>
>>help!!
>>
>>
>>Amit,
>>
>>If the NSSA has multiple ABRs and there is no
>>forwarding
>>address, routers outside the NSSA will always go
>>through
>>the translating ABR independent of whether or not it
>>is the shortest path.
>>
>>Thanks,
>>Acee
>>
>>
>>
>>>Regards
>>>Amit
>>>
>>>
>>>--- "Pat Murphy - [phone]"
>>><[e-mail]> wrote:
>>>
>>>
>>>>Amit,
>>>>
>>>>
>>>>
>>>>>          I can understand the point mentioned
>>>>>
>>in
>>
>>>>>RFC 2328 about the forwarding address but in this
>>>>>
>>>>>
>>>>case
>>>>
>>>>
>>>>>I am not able to appreciate the MUST condition
>>>>>
>>the
>>
>>>>>RFC(with Exception when u start aggregating
>>>>>
>>>>>
>>>>multiple
>>>>
>>>>
>>>>>routes).
>>>>>
>>>>>
>>>>It is true that type 7 LSAs that are aggregated
>>>>
>>into
>>
>>>>a single
>>>>Type 5 LSA with a 0.0.0.0 forwarding address do
>>>>
>>not
>>
>>>>need a
>>>>computed forwarding address. But usually an NSSA's
>>>>ASBR has no
>>>>way of knowing whether or not its self-originated
>>>>Type-7 LSAs
>>>>are aggregated by the NSSA's ABRs during
>>>>translation. Without
>>>>providing configuration to the contrary, the
>>>>
>>NSSA's
>>
>>>>ASBRs MUST
>>>>assume their Type 7 LSAs (with the P-bit set) will
>>>>not be
>>>>aggregated during translation and therefore MUST
>>>>compute a
>>>>non-zero forwarding address.
>>>>
>>>>I don't see much value in complicating the NSSA
>>>>
>>spec
>>
>>>>by
>>>>softening this requirement for Type 7 range
>>>>aggregations or
>>>>for the case of an NSSA with just a single ABR.
>>>>
>>Its
>>
>>>>hard to
>>>>imagine any IPv4 application where the forwarding
>>>>address
>>>>cannot be set (IPv6ers care to jump in???). Note
>>>>that when the
>>>>P-bit is clear, the new NSSA spec does allow a
>>>>
>>Type
>>
>>>>7 LSA to
>>>>have a 0.0.0.0 forwarding address.
>>>>
>>>>
>>>>
>>>>>An ABR for an NSSA does not generate type 4
>>>>>
>>>>>
>>>>summaries for
>>>>
>>>>
>>>>>ASBRs within the NSSA.
>>>>>
>>>>>
>>>>This is because all type 5 LSA translations of
>>>>
>>type
>>
>>>>7 LSAs are
>>>>originated by NSSA ABRs, not NSSA ASBRs.
>>>>
>>>>
>>>>
>>>>>Hence a forwarding address is
>>>>>required to compute the best path to the prefix
>>>>>advertised in the translated LSA.
>>>>>
>>>>>
>>>>If the NSSA's ABR translators used a 0.0.0.0
>>>>forwarding
>>>>address in their Type 5 LSA translations, these
>>>>
>>Type
>>
>>>>5 LSA
>>>>translations would cause packets to be forwarded
>>>>directly
>>>>through their originating ABR rather than via a
>>>>potentially
>>>>more preferred path through a different NSSA ABR.
>>>>Note that
>>>>this is always the case when type 7 ranges
>>>>
>>aggregate
>>
>>>>Type 7
>>>>LSAs into a Type 5 LSA with a 0.0.0.0 forwarding
>>>>address.
>>>>
>>>>Pat
>>>>
>>>>
>>>
>>>__________________________________________________
>>>Do You Yahoo!?
>>>HotJobs - Search Thousands of New Jobs
>>>http://www.hotjobs.com
>>>
>>>
>>>
>>
>>--
>>Acee
>>
>
>
> __________________________________________________
> Do You Yahoo!?
> HotJobs - Search Thousands of New Jobs
> http://www.hotjobs.com
>
>


--
Acee