On Sun, 2002-11-24 at 09:18, Kurt Erik Lindqvist wrote:
> > Absolutely agree. I've experienced the both the VPN and network 10
> > addressing situation concurrently with IPv4, in addition to having to
> > come up with bodgey solutions, I spent two months just saying to my 
> > self
> > "customers should just get registered address space, and make my life
> > (and theirs') a whole lot easier."
> 
> Good point!
> 
> >
> > Globally unique site locals would fix a lot of the issues I had with
> > trying to work out how I might use site-locals in large IPv6 network.
> >
> 
> Uhm, why not go with what you proposed above instead?
> 

As I understand it, the goal we are trying to achieve with site-locals
(whatever model you follow), is to have locally controlled and assigned
address space, independent of the global address space your provider
gives you.

I agree, if your global address space assignment from your provider is 
stable enough, and you don't have a requirement for addressing
independence, you could simplify your network even further, and not use
site-locals in any form at all. 

Using site-local and global addressing concurrently provides separation
of internal verses external connectivity, which then allows external
connectivity (and the associated global prefixes) to be changed, with
(or at least this appears to be the goal) no impact on internal
connectivity.

What I think the new (pseudo?)globally unique site-local addressing
models are doing is extending the scope of internal communications from
a site (geographical or otherwise), to potentially global ie. they are
saying that the potential for internal communications may be global, so
let's create a separate address space with global uniqueness, but only
for use with internal communications.

For actual external communications, use existing publicly routable
global addressing.

Once all the organisations connected to the IPv6 Internet have globally
unique site-local addresses, an illogical progression would be for them
all to connect themselves up via backdoor connections - to the point
where the actual Internet could be switched off :-)

However, they will face the same route table explosion / aggregation
problems that the real Internet already has, as well as the associated
security problems, and will probably leave global connectivity up to the
people who make it their business to do it properly - ISPs. 

Mark.


--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to