Re: Addr: Local Preference Computation

[Curtis Villamizar <[e-mail]>]

In message <[e-mail]>ge>, Joel Halpern writes:
> I have been reading the local-pref/MED discussion, and trying to tie it
> back to my understanding of the assumptions that had been made originally.
> 
> As I understood things, the point of "local-pref" was to ensure that all
> border routers of an AS made the same choice of path for a given destination.
>  
> This would then allow the appropriate BGP advertisement to be generated by
> all AS boundary routers.
> 
> If I am reading the discussion right, the routing loops occur because
> the local-pref value is not sthe only thing being used.  Therefore, 
> internal routers (who do not have the full policy of the external border
> routers) are getting confused about what is happening?
> 
> 1) I would strongly suggest that the discussion would be helped by an 
>     explicit statement in each proposal of how the fields (MED an and 
>     local-pref) are to generated and compared for route selection.
> 2) If the old assumptions are no longer applicable, I would appreciate
>     an explanation of why.  The most obvious cause would be that some
>     underlying precept turns out not to be the case.  For example, the
>     local-pref stuff is basically based on assuming similarity of policy
>     mechanisms on all border routers of an AS. Does that no longer obtain?
> 
> Yours, with apologies for not being as current on some of the thinging as
> I should be,
> Joel M. Halpern			[e-mail]
> Newbridge Networks Inc.

MED has been used for a long time to pick the exit point that leads to
the shortest path in a peer's topology.  This was done by setting
outbound MED according to IGP cost and passing the MED learned from
EBGP into IBGP.

Our normal use is to give a unique local-pref to each border AS for a
given prefix.  In the event that some of the 30,000+ prefixes have two
border AS providing the same local-pref, I'm not fussy about which way
things go out as long as a routing loop doesn't form.

Other providers routinely set a fixed local-pref per all other ISPs
and a fixed local-pref for all direct customers.  This results in a
lot of local-prefs being the same.  It was for this reason that
aspath-length was introduced to route selection in Cisco's
implementation, regardless of what the spec said.  Even so, equal
as-path lengths could lead to resolving the decision at MED or lower
with two AS providing MED.

I think Rich Woundy's "MED election" is a good way to solve this.  If
there is agreement, we simply need to clearly write this up so vendors
all use the same algorithm and end up selecting the same route and
then put the text into BGP4.

The only reason I suggested comparing MED across AS is because I knew
it was trivial to code while "MED election" takes a non-zero amount of
effort.  MED election is a cleaner way to eliminate the route loops.
(In a nutshell, "MED election" is compare all routes from the same AS
first using MED, then compare the best from each AS ignoring MED.  If
I got this wrong, all the more reason to put it in BGP4 so someone
writing BGP4 code doesn't).

Curtis