From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Sun Apr 20 2003 - 23:39:48 GMT-3
At 5:21 PM -0700 4/20/03, Umair Hoodbhoy wrote:
>What if you wanted to test this on the fly and not wait for historical
>stats?
>
>-- Umair
The reality is that a provider doesn't have to honor your MEDs.
Typically, anything beyond a basic routing policy is contractually
negotiated between provider and subscriber.
So, my first real-world approach would simply to call the provider
and ask them. Don't be surprised if they immediately tell you they
don't honor the MED. MEDs are valuable, but can also be
trouble-prone in large networks. See
http://www.isi.edu/in-notes/rfc3345.txt as well as the Cisco
deterministic MED discussions -- there are some good tech notes.
What I find more common, when providers will let you set up exit
preferences, is to set mutually agreed communities.
While it may not come across in the lab, one of the things that
really has to be grokked about global routing is that carriers can be
depended to handle your traffic a certain way if, and only if, you
pay them to do it that way. Sometimes.
This archive was generated by hypermail 2.1.4 : Thu May 01 2003 - 13:35:59 GMT-3