[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Aggregation Implies Provider Dependence // Re: [RRG] ALT's strong aggregation often leads to *very* long paths



Christian Vogt wrote:
Robin,

I agree with you that ALT's aggregation of provider-independent (hence
expectedly very scattered) EID space leads to longer-than-optimal paths.
I would like to address another adverse side-effect of aggregation:

ALT-style aggregation IMO violates the design objective of provider
independence.  It binds an edge network owning a piece of provider-
independent EID space to a particular lowest-level router in the ALT
hierarchy.  The edge network thereby gets tied to the "ALT provider"
operating this router, very similar to how edge networks with provider-
allocated address space are nowadays tied to their (physical) provider.

As a result, while address mapping solutions (such as Six/One, APT,
LISP) yield independence from physical providers, ALT introduces a new
dependence on "ALT providers".

We discussed this at the IST-RING workshop, as you may recall, Christian, and I believe that it was Jari who coined the term at the RRG meeting. While it is true in the abstract that you might have some dependence on your ALT provider, and while it is also true in the abstract that if you wish to change ALT providers, you may be forced to renumber, I think there are a number of mitigating possibilities here.

First of all, I could easily imagine partnerships forming between ALT providers that share portions of the address space and allow their customers to renumber from one partner to another. This isn't pure independence, but it is something.

Second, the degenerate scenario that you describe is that we are fully disaggregated. That means that there is a lot of STATE. It says nothing about RATE. This will depend very much on how the LAT is managed. If the LAT is extended to the last mile link without any static announcements, we would have precisely what we have today. However, if LAT announcements are relatively static, then the question is this: who cares?

Eliot

--
to unsubscribe send a message to rrg-request@psg.com 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