anything that encourages multi-faced DNS is a bad idea.

Keith

> At 02:43 AM 10/28/02, Markku Savela wrote:
> 
> > > When a multi-sited implementation gets site-local addresses from
> > > the DNS (assuming that it runs two-faced DNS and returns site-locals),
> > > how will the multi-sited host know which site the addresses are in?
> >
> >My guess, and tentative implementation is that the resolver library
> >completes the scope id to the address using the interface from which
> >the DNS reply came in.
> 
> I keep winding-up back here, myself, but it isn't really a satisfactory
> answer for several reasons:
> 
>          (1) A multi-sited node might receive different sets of
>                  addresses in response to the same query, based on
>                  which site's DNS server was tried first.  There is
>                  no provision to try another DNS server if the first
>                  one succeeds, so there is no way to be sure you've
>                  gotten all of the available addresses to reach a
>                  node.
> 
>          (2) This solution presumes some relationship between the
>                  routing hierarchy and the organization of DNS servers.
>                  For example, it assumes that I will use the DNS
>                  server that is configured with two-faced DNS for my
>                  site.
> 
>          (3) The two-faced DNS solution requires that we store routing
>                  information in a DNS configuration file.  This creates
>                  another way for the network to break when routing
>                  information changes.
> 
>          (4) If the DNS requests are done via global addresses, I am
>                  not sure that it is safe to assume that the results
>                  will always come back through the correct interface.
>                  What if there were a partition in Site1, but parts of
>                  Site1 could still reach other parts of Site1 using
>                  global addresses, by sending across the global
>                  Internet.  Is that possible?
> 
> Anyway, we'd certainly have to document the "two-faced" DNS concept
> and describe this solution in more detail before I'd feel comfortable
> counting on it to work.
> 
> Margaret
> 
> 
> --------------------------------------------------------------------
> 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]
> --------------------------------------------------------------------
--------------------------------------------------------------------
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