Re: Indirect redistribution

From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Sun Jul 04 2004 - 18:49:36 GMT-3


At 4:35 PM -0700 7/4/04, <alsontra@hotmail.com> wrote:
>Yes, this is what I was asking. I would have thought that protocols
>redistributed information from their individual topology databases. The idea
>that redistributed information comes from the RIB is very new to me.
>However, it clearly explains why single device indirect redistribution is
>not possible.
>Just to be sure that I'm clear, the RIB takes the best (lowest admin
>distance, longest match, etc) routes from all protocols and it is from this
>information or database that mutual redistribution occurs.

     Yes. Order of preference:

     1. Completely new route: INSTALL
     2. More specific of existing route: INSTALL
     3. Lower administrative distance: REPLACE
        Equal administrative distance
          If STATIC INSTALL & LOAD BALANCE*
           Else
             If LOWER METRIC REPLACE
               Else
                if EQUAL METRIC INSTALL & LOAD BALANCE*

*up to maximum paths.
** BGP first checks its own Loc-RIB

>??? Simply put,
>the RIB's singular selection of "most preferable" routes can cause the lose
>of less preferable protocol information.

Yes. It's also a safety measure for loop prevention to only advertise
routes that the local router actually is using for forwarding.

>As I understand it, the solution to this problem is to mutual redistribute
>all protocols. ??? Correct?

Could you restate the problem? This is for lab practice, I assume --
in the real world, you want to redistribute as little as possible.



This archive was generated by hypermail 2.1.4 : Sun Aug 01 2004 - 10:11:46 GMT-3