Re: static routes, dynamic LAN advertisements, and RIPv2

[Paul Ferguson <[e-mail]>]

At 01:54 PM 1/30/96 -0500, from cais.com wrote:

>
>But, on the other hand,  to my knowledge, there is not a stochastic
>study on record that models the network performance of a hierarchical
>routed networks in a highly asymmetric hierarchy; and the abuse of
>the /N prefix feature lends to asymmetry.
>
>The Kleinrock paper is fairly clear in stating that symmetry is a
>major assumption in his routing model.  The use of "classless
>prefixes" and masks at the network engineering level has the potential
>to impact the symmetry approximation of the cluster model.
>
>The supporting work in the field for asymmetric hierarchical routing
>is practically non-existent due to the complexity of the stochastic
>model and the historical lack of commercial interest.
>


Funny you should mention the inherent problems of asymmetric routing.

I assume you've read Enke Chen & Tony Bates drafts:
 
 Current Practice of Implementing Symmetric Routing and Load Sharing in 
 the Multi-Provider Internet <draft-ietf-idr-symm-multi-prov-02.txt>

 Application of the BGP Destination Preference Attribute in Implementing 
 Symmetric Routing <draft-ietf-idr-dpa-application-02.txt>

 Destination Preference Attribute for BGP <draft-ietf-idr-bgp-dpa-04.txt>


Not a panacea, <draft-ietf-idr-bgp-dpa-04.txt> clearly states that:

   The AS that sets this attribute must include its AS number in the
   attribute.  A BGP speaker may use the "LOCAL_PREF" attribute to
   select a different path other than the one specified by the DPA
   attribute value.  This does not preclude an AS from re-setting this
   attribute.  However, coordination with the upstream and/or downstream
   neighbors is strongly recommended.


- paul