Re: Detecting inactive neighbors over demand circuit

[Erblichs <erblichs at EARTHLINK.NET>]

Anu,

        Assuming that the Router LSA needs to be flooded
        in the near future (Not a DNA LSA), what is the harm
        to having it being flooded earlier?

        Assuming that a numbers of routers were
        booted at the same time, to minimize sychronization
        of flooding updates (assuming this is considered
        a problem), flooding a LSA earlier makes some
        sense to spread out the flooding.

        Assuming that the initial origination of the
        LSA is a active probe, the flooding to other
        routers florcefully verifies that all links
        are still active at this one point in time...

        Assuming that a non-floodable LSA is originated,
        the router LSA still needs to be flooded and
        not all the links are known to be in the same
        state at the same time..

        Lastly, if a scan identified a checksum corrupted LSA,
        this pre-flood could/would resych the LSDB at a
        earlier point in time. Could, because some routers
        would just go down into maintainance mode..

        Mitchell Erblich
        ----------------


armstrong mathiayalagan wrote:
>
> Hi,
>         This draft recommends (May clause) Router LSA to be flooded for Neighbor
> probing.
>         With respect to the applicability of this draft to OSPFv3, I feel Link LSA
> is a good choice for using in neighbor probing than Router LSA.
>         In Demand Circuit, there are more chances for having LSAs of different
> sequence number at both ends.
>         If neighbor probing uses Router LSA, if the Router LSA is having higher
> sequence number than the neighbors database, then the neighbor will flood
> these LSA in all the non demand circuit interface. This can be avoided by
> using Link LSA.
>         Comments are welcome.
>
> Regards,
> Anu
> --
> ___________________________________________________________
> Sign-up for Ads Free at Mail.com
> http://promo.mail.com/adsfreejump.htm