Scott Brim wrote:
My thinking was: If a site (i.e. an area of the network addressed only
by EIDs and surrounded by xTRs) is connected to more than one ISP/AS,
and the how can you
- have the xTRs in PEs, under control of the ISPs
- have good cooperation between the xTRs, and
- not depend on cooperation between the ISPs
Hi Scott,
Let me use APT as an example of how this is possible.
In APT, a site provides preferences and weights for TE separately to
each of their providers. (Let's call these preferences and weights "TE
data".) So there is an exchange of information between customer and
provider, which would need to occur in order to establish a business
relationship, anyway. But there is no need for the TRs to cooperate
directly.
TE data is propagated separately from each provider to all default
mappers in the network. Default mappers store all TE data along with the
corresponding ETR address. When an ITR needs to find an ETR address for
a site, it can ask its default mapper. The default mapper assigns an ETR
address to the ITR based on the accumulated TE data.
In the grander scheme of things, the major difference here (versus LISP)
is not related to the CE vs. PE debate. The difference is that we
require TE data to be the same globally -- that is, we don't support
source-specific TE (from the ETR's perspective). This is simply because
we have picked a different point in the tradeoff space -- one, global
set of TE data allows for one, global system for distributing it. To
summarize, I believe source-specific TE and our hybrid push-pull model
are mutually exclusive.
-Michael
--
to unsubscribe send a message to [EMAIL PROTECTED] with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg
