[c-nsp] BGP Balanced

[Ben Crocker]

To clarify here's a couple of links I had to show how J & C select  
routes (the newest ones I could find last time I had to check  
something).

http://www.cisco.com/en/US/partner/tech/tk365/tk80/ 
technologies_tech_note09186a0080094431.shtml
&
http://www.juniper.net/techpubs/software/junos/junos64/swconfig64- 
routing/html/protocols-overview4.html#1014020

Notice they aren't exactly the same.

I found this book to be very helpfull as well.

BGP 4 Inter-Domain Routing in the Internet
John W Stewart III
ISBN# 0-201-37951-1

Perhaps it can help with the MED misunderstandings.

Laters

Ben


On Aug 27, 2004, at 8:58 PM, Hudson Delbert J Contr 61 CS/SCBN wrote:

> ejay,
>
> why is 'med' so 'mis-understood'?
>
> it is a metric seting, plain and simple.
>
> it is used to when AS ZZZ and AS YYY have multi-links w/each other.
>
> it is used to set by ZZZ to inform YYY ebgp speaker of what ZZZ's
> preferred entrance points are and is retained by YYY just in case
> the preferred ZZZ POE is down upon re-entrance.
>
> the default is usually = 0.
>
> it is only of local to the sharing AS's as this info is reset=0 if the
> traffic is
> passed to a third AS.
>
> the point of 'med' assignment is in route maps.
>
> it is usally step #8 in BGPv4 Path Selection..
>
> algorithm (i think) for path selection:
> 1=	next_hop is reachable and no others exist
> 2=	largest wgt.
> 3=	if no wgt supplied or all wgts =, use larget lcl pref.
> 4=	if local pref =, path .eq. route oroginated by 0.0.0.0 in next_hop.'
> (whew!! had to really thing about that one)
> 5=	shortest AS Path
> 6=	if all path = len, use lowest orig code.
> 7=	if orig codes =, use lowest med, remember def=0 usually.
> 8=	if all med =, choose ext over int path.
> 9=	if path = internal, use path thru closest interior routing protocol
> nbr
> 	using local routing protocol. (???? not sure aboutthis one)
> 10=	use lowest BGP router id, loopback always a pretty good choice.
>
>
> v/r,
> piranha
>
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net]On Behalf Of Ejay Hire
> Sent: Thursday, August 26, 2004 3:05 PM
> To: bep at whack.org; 'Alexandra Alvarado'
> Cc: cisco-nsp at puck.nether.net
> Subject: RE: [c-nsp] BGP Balanced
>
>
> On an unrelated note, MED is poorly implemented and
> understood on the internet at large, and should usually be
> avoided.  Misconfiguration can cause route flapping and is
> difficult to troubleshoot.
>
> -ejay
>
>> -----Original Message-----
>> From: cisco-nsp-bounces at puck.nether.net
>> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of
> Bruce Pinsky
>> Sent: Tuesday, August 24, 2004 7:03 PM
>> To: Alexandra Alvarado
>> Cc: cisco-nsp at puck.nether.net
>> Subject: Re: [c-nsp] BGP Balanced
>>
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Alexandra Alvarado wrote:
>>
>> | What about BGP Multi Exit Discriminator (MED)?
>> |
>>
>>
>> Generally speaking, that would only apply if you actually
> had
>> multiple BGP
>> peers to the same provider and therefore could choose
> between
>> exit points.
>> ~ Here it sounds like you will simply have one exit path
> and
>> need to send
>> traffic over multiple physical connections.
>>
>> As some have pointed out, BGP may be overkill in this
> situation.
>>
>> - --
>> =========
>> bep
>>
>> -----BEGIN PGP SIGNATURE-----
>> Version: GnuPG v1.2.2 (MingW32)
>>
>>
> iD8DBQFBK9c/E1XcgMgrtyYRAtaCAJ9vFLiBoG5ZqymRCrbCQWPJMDeS7ACe
> J+3T
>> 91a4ILKyIGO7mVP/RyDMnsw=
>> =0hEi
>> -----END PGP SIGNATURE-----
>> _______________________________________________
>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
>