[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
re: [RRG] yetAnotherProposal: AS-number forwarding
> -----邮件原件-----
> 发件人: Peter Sherbin [mailto:pesherb@yahoo.com]
> 发送时间: 2008年3月10日 14:20
> 收件人: Xu Xiaohu; 'Lixia Zhang'; 'Lars Westberg'
> 抄送: rrg@psg.com
> 主题: re: [RRG] yetAnotherProposal: AS-number forwarding
>
> > The last approach is to introduce an independent host id namespace and uses
> > a whole GIRO address as locator.
> >
> > Which one do you prefer? Or you have any other better idea?
>
> ID/Loc split comes in various forms over and over again with people seem to
> concentrate on a soft transition of the current system into a scalable one with
> a
> minimal disruption. As prudent as it sounds it might not yield anticipated
> benefits.
> More desirable approach is to abandon the current Internet entirely and build
> a new
> one from two independent, separately managed pools of id: a locator pool and
> a name
> pool, where names move freely around a rigid hierarchy of locators.
I basically agree with your idea of two independent, separately managed pools of id and locator namespace, but I don't like the idea of abandoning the current Internet entirely.
The IPv4 address can still be used well as a local locator, not global locator, with an appendix of locator domain (LD) prefix/ID. For example, there are multiple LDs in the world, each LD can use independent IPv4 address space, different LD only exchange LD prefix/ID information with each other, in this way, there are at least three benefits: 1) the address depletion issue is solved; 2) the route stability is improved greatly since the route churn in one LD will not be flooded into other LDs and the LD-prefix can even be aggregated further; 3) the deployment cost is relatively low since only the LD border routers need to be upgraded and the most internal routers within LD can still be IPv4-enabled router. The LD prefix/ID is strictly hierarchical and topology-dependant, or even geographical-location related (see GIRO). This is some idea of HRA.
IMO, the strict and full hierarchy in locator namespace is not a good idea as it will limit the flexibility of address assignment and the efficiency of address utilization.
Best wishes,
Xiaohu XU
> > > -----�ʼ�Ô��-----
> > > ������: owner-rrg@psg.com [mailto:owner-rrg@psg.com]
> ����
> Lixia Zhang
> > > ����ʱ��: 2008��3��10�� 12:22
> > > �ռ���: Lars Westberg
> > > ����: rrg@psg.com
> > > ����: Re: [RRG] yetAnotherProposal: AS-number forwarding
> > >
> > >
> > > On Mar 9, 2008, at 11:40 PM, Lars Westberg wrote:
> > >
> > > > Hi,
> > > > I haven't had time to make a draft but I think it make sense for the
> > > > discussion. However, I don't know if it already have been discussed
> > > > so....
> > > >
> > > > The proposal are simple: re-use AS-numbers into the forwarding of
> > > > packets such that prefixes could be aggregated per AS. One simple
> > > > implemetation is that the packets are tunneled and that the tunnel-
> > > > address is associated to a AS-number. The AS-numbers can be assigned
> > > > to the IP-addresses by DNS or by define a small address-prefix to AS-
> > > > numbers.
> > > >
> > > > Comments?
> > > >
> > >
> > > in an ideal world, yes having AS number as part of address used for
> > > routing has great benefit. see the slide from a talk in 2006
> > > (http://www.cs.ucla.edu/~lixia/0612Australia.pdf
> > > , but ignoring the title), slide 17 & 18 is about this.
> > > If we had a chance to influence address structure, you'd want to
> > > include other info in addition to AS (as large ASes span large areas,
> > > TE would want more info to do better job).
> > > we have another paper (http://www.cs.ucla.edu/~rveloso/papers/
> > > giro.pdf) showing the benefit for including location info (which
> > > should be a subfield after AS number)
> >
> > With support of routing aggregation at any desired level, I can't image that
> > GIRO alone can do much help in routing scalability issue unless it brings
> in
> > some id/locator split idea, like GSE, SIX/ONE or HRA.
> >
> > The first approach is to split the GIRO address into two parts: id and
> > locator. In this approach, IPv4 address part in GIRO address is used as id,
> > which means there is no need to change the hosts however it doesn't address
> > the IPv4 address depletion issue. From this perspective, it's much like eFIT
> > or LISP but it introduces a new locator space.
> >
> > The second approach is to make hosts to negotiate the bunch of several GIRO
> > addresses.
> >
> > The last approach is to introduce an independent host id namespace and uses
> > a whole GIRO address as locator.
> >
> > Which one do you prefer? Or you have any other better idea?
> >
> > Best wishes,
> > Xiaohu XU
> >
> >
> >
> > --
> > 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
> >
>
>
>
>
> __________________________________________________________________________
> __________
> Be a better friend, newshound, and
> know-it-all with Yahoo! Mobile. Try it now.
> http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ
--
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