[isp-bgp] Re: MEDs from different ASes

[Ton Schoenmakers <ton-s-lijst@...>]

From the Cisco website....

There are two Border Gateway Protocol (BGP) configuration commands that can influence the Multi Exit Discriminator (MED)-based path selection, the bgp deterministic-med and the bgp always-compare-med commands.

Enabling the bgp deterministic-med command ensures the comparison of the MED variable when choosing routes advertised by different peers in the same autonomous system (AS). Enabling the bgp always-compare-med command ensures the comparison of the MED for paths from neighbors in different autonomous systems. The bgp always-compare-med command is useful when multiple service providers or enterprises agree on a uniform policy for setting MED. Thus, for network X, if Internet Service Provider (ISP) A sets the MED to 10, and ISP B sets the MED to 20, then both ISPs agree that ISP A has the better performing path to X.

There is sometimes confusion between the two commands. This Tech Note explains the differences in how the bgp deterministic-med and bgp always-compare-med commands can affect MED-based path selection and how each command changes the behavior of BGP when choosing a best route.

Note: These two commands are not enabled by default. Also, the two commands are separate, enabling one does not automatically enable the other.


Full description at http://www.cisco.com/en/US/partner/tech/tk365/tk80/technologies_tech_note09186a0080094925.shtml


We used it in a dual-ISP dual-circuit environment when we had plenty of money to burn..
Nowadays we traded in most of the dual-circuits for a single RouteScience box.

Ton Schoenmakers



Tulip Rasputin wrote:
> Hello,
>
> I am able to comprehend the concept of comparing MEDs for the routes that we
> recieve from the same AS. What confuses me is when, would we really want to
> compare MEDs across different ASes? MEDs can mean different things to
> different operators, one could be reflecting the internal IGP costs in the
> MED, the other could be doing something totally different. How and in what
> scenario, will we want to compare MEDs across different ASes?
>
> Can anybody throw any light on this? Has anybody ever done that? And if yes,
> then when?
>
> One reason which i can think of at the top of my head is to avoid RFC 3345
> oscillations. We may use always-compare-MED to prevent those. But then there
> are many topological solutions that can help prevent those oscillations.
>
> Thanks,
> Rasputin







To unsubscribe via postal mail, please contact us at:
Jupitermedia Corp.
Attn: Discussion List Management
475 Park Avenue South
New York, NY 10016

Please include the email address which you have been contacted with.