From: gscisco@xs4all.nl Cisco (gscisco@xs4all.nl)
Date: Mon Apr 24 2006 - 15:23:58 GMT-3
Scott,
Thanks. Obvious indeed. It's funny since with the first solution by
prepending I didn't prepend even add an AS to one of the BGP advertising
routers because of the AS path. I think I must have gotten confused when
playing with the MED.
Anyway added a prepended AS to the redistributing router and then it does
start to look at MED.
Thanks again for making it clear :)
Vincent
On 4/23/06, Scott Morris <swm@emanon.com> wrote:
>
> Per my experience, that document is actually correct. You have other
> issues
> though! BGP paths (at step 7) are chosen in numerical order of origin code
> (0 = internal (IGP), 1 = external (EGP), 2 = Incomplete/Unknown) (Step 5
> occurs first, which would indicate eBGP neighbor)
>
> Now, bear in mind that you say the route was redistributed... In the
> Cisco
> world, that will set the origin code as incomplete. This is irregardless
> of
> the iBGP or eBGP relationship (as this type code doesn't change in
> updates).
>
> The preference you are seeing is likely due to other information such as
> the
> router-id of the advertising neighbor being lower. Looking through the
> RFC,
> the spec lays this out as one of the tie-breaking mechanisms. On the
> other
> hand, I'd probably think the decision would take place even before that
> based on AS-path length! If your PE's are the same AS, then that's a damn
> short AS path. :) If you learn it from the CE, then your path would be
> longer, and this is done in step 4!!!
>
> So you aren't making it to step 7 in preferring your internal path to the
> other PE!
>
> HTH,
>
>
> Scott Morris, CCIE4 (R&S/ISP-Dial/Security/Service Provider) #4713, JNCIE
> #153, CISSP, et al.
> CCSI/JNCI
> IPExpert CCIE Program Manager
> IPExpert Sr. Technical Instructor
> smorris@ipexpert.com
> http://www.ipexpert.com
>
>
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> gscisco@xs4all.nl Cisco
> Sent: Sunday, April 23, 2006 5:00 PM
> To: ccielab@groupstudy.com
> Subject: BGP IBGP vs EBGP path
>
> Hi,
>
> I hope you guys can help me understand this as IBGP and EBGP routes behave
> different than I have always understood them.
>
> From what I read in cisco's "BGP Best Path Selection Algorithm" -
>
>
http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094431
> .shtml#background
>
> 7 - Prefer eBGP over iBGP paths.
>
> However it is not the behavior I'm seeing myself
>
> I have the following setup
>
> CE1 === default route -> / <- static redistributed in BGP === PE 1
>
> CE2 ====== running BGP ====================================== PE 2
>
>
> PE1 and PE2 are within the same AS and run iBGP between them.
>
> I have a network 157.161.77.0/24 which PE1 has a static route for which it
> redistributes into BGP. PE2 gets this same network from CE2 via (e)BGP. It
> however does not take this as the best BGP path as keeps prefering the
> internally learned one via iBGP
>
> PE3#sh ip b 157.161.77.0
> BGP routing table entry for 157.161.77.0/24, version 5
> Paths: (2 available, best #2, table Default-IP-Routing-Table)
> Advertised to non peer-group peers:
> 161.239.99.2
> Local
> 172.16.1.3 (metric 11) from 172.16.1.3 (213.239.99.3)
> Origin incomplete, metric 10, localpref 100, valid, internal, best
> <<<<< iBGP learned path
> 65000
> 161.239.99.2 from 161.239.99.2 (161.239.99.2)
> Origin IGP, localpref 100, valid, external <<<<<<<<<< eBGP
> learned
> path
>
>
> Am I misinterpreting cisco's path algorithm ? Furthermore I also see that
> it
> should have taken the the latter path as it's Origin is better as well
> (IGP
> goes before "incomplete") ?
>
> Can anyone shine a light on this for me as I'm quite confused.
>
> Thanks in advance,
>
> Vincent
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Mon May 01 2006 - 11:41:59 GMT-3