Re: [RRG] Sceptically on compact interdomain routing (Modified by Victor Grishchenko)

[Victor Grishchenko <gritzko@post.dcn.ru>]

On 18.10.2005, at 20:11, Joel M. Halpern wrote:
> While theoretically hierarchical routing uses logarithmic space, there 
> are several ways in which that turns out not to be the case:
> 1) Even with best case address allocation, we will probably have one 
> entry per top level AS.  For a sense of scale, we will exhaust toe 16 
> bit AS number space in the near future.
> 2) In practice there are far more than 1 prefix per AS.  This occurs 
> for many reasons.  The reason thet folks like Geof Huston have 
> indicated is largest is multi-homing.
OK. But, there are several ways in which that turns out to be radically 
better.
E.g. RFC2374 proposes IX-based aggregation for IPv6 (i.e. IP numbering 
roots are IXes).
So, the number of roots is rather limited (some other trick are 
possible to reduce that number even further).

> Hence, a rough analysis indicates that if the compact routing research 
> can be made practical (which is a major open question) then even if 
> IDs were assigned to ever site on the net, we would still end up with 
> a smaller table than we have now.  That is a pretty good result.
It will remain to be another variant of splitted Dijkstra solution (see 
paper), i.e. there will be no difference in order, just a coefficient. 
(If I get your point right. There is also another way -- to apply CR on 
top of ASes. This second way dramatically reduces core routing tables 
but leads to further path inflation plus some other considerations.)

> But I certainly am not asking not to continue investigating just 
> because there are more hard problems further out.
Agree.

> PS: I am not a researcher with an account on arxiv, and it appears 
> that accounts are intended for folks who will be submitting papers.
Sorry, I submitted it today. It takes some time to propagate to 
different mirrors, etc.
I think, this link will work tomorrow (after 00:00 GMT),
		http://arxiv.org/abs/cs.NI/0510052

--

	Sincerely,

			Victor


--
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