Le 25-juil.-08 à 03:58, Dan Jen a écrit :
details.
Thanks for the pointer.
Here I'm confused. In the previous paragraph you claim the contrary.
Really? This paragraph claims that PA prefixes are aggregatable by
network providers in the core, and once these edge prefixes are
aggregated, reachability status of these edges no longer result in
updates to the global routing table, leaving reachability tracking to
the transport layer. I don't see where we claim the contrary. Can you
show us?
Transport protocols
will track the status of individual paths and make adjustment
accordingly.
Here you say transport protocols will track status of individual paths....
While previously you state...
Besides
potentially improved performance, the simultaneous use of multiple
paths easily tolerates the loss of any single path, hence it becomes
less important to keep track the status of any individual paths.
.. and this statement sounds like you do not need to keep track of individual paths.
Ahh, we should probably reword this a bit. Our point is that under the
map&encap approach, we remove the PI prefixes from the core, but they
still exist, as they are used by the edge sites. Since they still
exist, we call it separation rather than elimination. This is in
contrast to the transport scheme, where edge sites actually stop using
PI prefixes in exchange for PA prefixes from their providers. Thus the
PI prefix is eliminated from use.
Much more clearer in this way, thanks.
Cheers
Luigi
Sorry about the confusion. We'll try to clear that up in our next
revision.
Dan Jen