To:
ogud@ogud.com (Ólafur Guðmundsson)
Cc:
bmanning@ISI.EDU, johani@autonomica.se, dnsop@cafax.se
From:
Bill Manning <bmanning@ISI.EDU>
Date:
Mon, 7 Oct 2002 20:37:08 -0700 (PDT)
In-Reply-To:
<5.1.1.6.2.20021007223831.01713e90@localhost> from Ólafur Guðmundsson at "Oct 7, 2 10:45:12 pm"
Sender:
owner-dnsop@cafax.se
Subject:
Re: Interim signing of the root zone.
% At 14:10 2002-10-07, Bill Manning wrote: % > some concerns: % > % > DS only works in snapshot code. And the publicly availble % > snapshots have known, serious operational problems. We -REALLY- % > need more stable code before committing this to production. % % Bill this is version 00 of the draft, your concerns are noted but % this particular experiment is not starting next week or next month. % This is the documentation for the experiment and Johan is seeking feedback. documentation for -an- experiment. an experimental setup has existed and been running for over 2 years. Signing the root zone in this testbed has been operational for three months, nearly as long as ther has been DS capable code. I remain leary of experimentation with the live system. % > the selection of RIRs. RIRs -DO- have the DNS as a primary % > field of activity. (see in-addr.arpa.) The holders of % > forward space (.SE, DE, NL, etc.) become disinfranchised % > "customers". % % RIR are geographically competent operators for this experiment, % for future production Layer 9 will become involved. "geographically competent" - now there's a turn of phrase :) -IF- this is really an experiment, with the live system, then bounding the experiment is prudent. I'd be -very- leary of giving even the suggestion of "early-implementor" bias to one vector of the possible keyholder pool. % > "sufficient number" and "out-of-band" are concepts that % > really need some concrete recommendations. % % Yes, suggestions. More experimentation in the operational testbed may generate some empirical numbers/processes that work. Not convinced that running this on the live system is reasonable. % > key duration should be better fleshed out. Experiences from % > the existing testbed may be useful. % % yes, suggestions please, % as well as key length key set size etc, etc. More work within the existing testbed will generate such numbers that have some grounding in experience. % > key publication methods have been explored but do need further % > work. % % agreed, this is one of many the research programs that this experiment % will hopefully shed some light on what works and what does not. % Issues involve: % - DS or KEY as published record % - where to publish % - is there a way to auto-configure resolvers trusted keying list So... why are we considering experimenting with the live, production root system at this time? IMHO, this is lunacy. We have a working, experimental system in play where most (all) of these issues can be tested. Folks that have serious commercial interests in a stable system will not be amused when we start experimenting with the systems that they depend on. % Olafur -- --bill