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

Re: A strawman analysis on locator identifiers vs. non-locator identifiers (was Re: A roadmap for end-point identifiers?)


You're simply wrong.  The scenarios I described are things that
applications do today using IP addresses as endpoint identifiers,
extrapolated to the world where DNS names are used instead.

I won't claim that it's impossible to use DNS names for endpoint
identifiers.  I do claim, and have explained in detail, why:

- using existing DNS names is a bad idea (due to overloading)
- using the DNS lookup system is a bad idea (due to performance issues)

You have the habit of dismissing anything that you don't understand as
irrelevant.  This is just another example.  Frankly it's getting old.


Dave Crocker said:
> Keith,
>>> KM>     * Ambiguous and overloaded. Quite often, a DNS name doesn't
>>> refer to
>>> KM>     a host.
>>> So?  What problem does that cause?  Specifically, please provide a
>>> usage
>>> example that would cause a problem.
> KM> Name EXAMPLE.COM maps to hosts B and C.  Hosts B ausing addresses B1
> and C1,
> KM> respectively.  Host A looks up EXAMPLE.COM, gets addresses B1 and C1,
> chooses
> KM> B1, and starts talking to it.  The application now has some state that
> is only
> KM> accessible to host B.
> KM> For some reason the connection gets broken.  Maybe host B changes its
> KM> address from B1 to B2.  A needs to re-establish a connection to host B
> -
> KM> host C will not work because the application now relies on state that
> only
> KM> B can access.  EXAMPLE.COM does not name host B uniquely;
> I am guessing that you have not read the MAST proposal and have not been
> noting the details that a number of us are seeking about actual
> requirements
> for an endpoint identifier.. I'm suspecting you also are not familiar with
> the
> other proposals.
> So, yes, you have invented a scenario that uses domain names in a way that
> will not work.  Unfortunately, no one is proposing such a broken scenario.
> So let me rephrase the question more simply and more rigidly:
>      What is it that _requires_ specifying using domain names in a way
> that
>      causes problems?
> If that question is not clear enough, then try:
>      What is _inherent_ in domain names that prevents their use as
> endpoint
>      identifiers?  Inherent, not merely "possible".
> It is always possible to create a broken design.  Citing such an example
> is
> not overly helpful.
> However I really should thank you. Much of these discussions have been
> marked
> by attempts to make design choices based on very abstract issues, rather
> than
> on looking on plausible and probable actual usage. This makes much of the
> discussion fundamentally not grounded, both literally and figuratively.
> We now have quite a few concrete proposals and specifications. We should
> make
> a point of using these data points as concrete examples of the solution
> space.
> d/
> --
>  Dave Crocker <dcrocker-at-brandenburg-dot-com>
>  Brandenburg InternetWorking <www.brandenburg.com>
>  Sunnyvale, CA  USA <tel:+1.408.246.8253>

IETF IPv6 working group mailing list
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6