[OCCAID] External Traffic Engineering Comm's

James james at towardex.com
Tue Jan 18 08:34:00 EST 2005


Few EINTAP's who are multihomed to more than one OCCAID POP have
expressed the lack of "IGP derived MED values" originating from
30071.

Unfortunately we are not able to provide IGP-based MED based
cost to our downstreams, due to technical limit of the routing
software (modified Zebra/Quagga) in use.

But we now have a plausible alternative solution. We are now
beta testing "path origination communities" for multihomed
EINTAP's who wish to test this out and let us know how it works
out for you.

Following OCCAID POPs are participating in this setup:

Best path routes seen from ZOA - 30071:8001
Best path routes seen from ZBW - 30071:8002
Best path routes seen from ZNY - 30071:8003
Best path routes seen from ZFW - 30071:8005

Basically all of our regions where we receive v6 transits from.

If you are for example multihomed to ZBW and ZNY regional POPs,
you can apply a route-map that seeks "30071:8002" to your Boston
peering and set a higher local preference, then set regular
local preference for non-matching prefixes on that peer.
Likewise, you can set a route-map that seeks "30071:8003" to
your New Jersey peering and set a higher local preference, while
setting default 100 on the rest of the remaining prefixes.

Obviously these communities are provided for people with 
sufficient BGP clue to attack their multihomed inefficiencies
to OCCAID POPs.

If you have questions, let me know. If you are multihomed to
more than one of the participating POPs above, please test this
out and see how this works out for you in better improving
your routes.

This is a pretty hard/hackery kludge way to fix the problem but
sometimes we have to do workarounds that are not very disruptive
in network setup, when using cheap routing gears :)

Thanks!

-J


More information about the Occaid mailing list