Re: BGP-4 usage document (revised)

[Curtis Villamizar <[e-mail]>]

In message <[e-mail]>, [e-mail] write
s:
> 
> Let me give an example:  Assume that site X is given 10.0 through 10.127,
> and site Y is given 10.128 through 10.255.  Both sites X and Y are
> misconfigured.  Then the current implementations will aggregate these
> advertisements to net 10, and the routing system sees two
> advertisements, 10 via site X, and 10 via site Y.

Right.  I lost the original context in my posting here and started
rambling about proxy aggregation in general but regained consciousness
by the end of the message.  :-)

[...]

> > Back to the original point- I agree with the point that no transit
> > provider should be proxy aggregating without examining the
> > consequences and also exhausting attempts to get the originator to
> > aggregate in the first place.  Therefore the default behaviour will
> > break things in subtle ways.
> 
> Yes, but the default behaviour has its uses, as Paul and Dimitry
> pointed out.  The BGP-4 editors should probably relegate the offending
> paragraphs to a footnote, or the "changes from the previous version"
> section.
> 
> Kannan

We all seem to agree on this.

Curtis