---------------------------------------------------------------------- Date: 4 Aug 1996 02:53:25 -0700 From: Simon Higgs Subject: Re: Who is archiving the shared-tld list? At 1:19 AM +0100 8/3/96, Neil Readwin wrote: >Is anyone archiving the list? I cannot find a pointer to an archive >on the Web pages. Neil. _____S_i_m_o_n___H_i_g_g_s_________________H_i_g_g_s___A_m_e_r_i_c_a_____ ... "I'm fine - it's the others" ............. President/CEO ............ _____e-mail: ___________ ______ .. .. ---------------------------------------------------------------------- Date: 4 Aug 1996 06:32:42 -0700 From: Kent Crispin Subject: Re: WG charter timeline Michael Dillon allegedly said: > > On Fri, 2 Aug 1996, Kent Crispin wrote: > > > Mar 1997 > > Submit Shared Top Level Domain Draft > > This isn't really complex stuff here. I think this should be Oct 1996. > > > Mar 1997 > > Establishment of the .SHARED TLD for purposes of testing > > STLD proposals. > > Oct 1996 > > > Jun 1997 > > STLD RFC > > Apr 1997 I actually think the stuff is fairly complex. Chris's suggestion for the first two milestones splits the difference between Oct 96 and Mar 97 at Jan 97, so how about if I use his date and Apr 97 for the RFC? I also added a sentence about motivation. Please, more comments! stld-charter: DRAFT Shared Top Level Domains Working Group (STLDWG) Charter Chair(s) o TBD o (Someone from IANA?) Internet Area Director(s) o TBD Mailing List Information o General Discussion: shared-tld@higgs.net o To Subscribe: shared-tld-request@higgs.net o Archive: http://www.higgs.net/mail/lists/shared-tld/shared-tld-digest.html Description of Working Group The Shared Top Level Domains Working Group is concerned with the technical and logistic requirements of creating shared domain name registration databases, and the administration of delegated top level domains by multiple domain name registries. The motivation for this concern is to minimize centralized management of all components of the name space. The primary products of this WG are three: First the STLD Draft, and second, a test implemtation of the ideas using the .SHARED TLD as a test, and three, a revised STLD draft incorporating the lessons from the experiment, which should become an RFC The areas of concern include o fostering an appropriate blend of competition and cooperation between cohort registries o the relationship between STLD registries and name servers o technical issues regarding management of the various distribute databases o the adminstrative procedures involved in running a registry serving a Shared TLD o etc... [help!!!] Goals and Milestones Jan 1997 Submit Shared Top Level Domain Draft Jan 1997 Establishment of the .SHARED TLD for purposes of testing STLD proposals. Apr 1997 STLD RFC - -- Kent Crispin kent@songbird.com "No reason to get excited", kc@llnl.gov the thief he kindly spoke... ---------------------------------------------------------------------- Date: 4 Aug 1996 07:23:49 -0700 From: Kent Crispin Subject: Re: Lightweight vs. heavyweight registries John R Levine allegedly said: > > >This is the good part about competition -- if Bogo-net was incompetent, > >then presumabley they won't remain in the business for long. :-) > > Ah, but what business? The registry business. > I'm expecting that the many if not most of these > outfits will be ISPs who are doing registry as a sideline to support their > customers, so they don't care all that much about the registry business. [snip] I don't think the concern you raise is really much of an issue for the following two reasons: 1) If they are doing registry service for their customers, and they do a bad job, presumably customers will go elsewhere. It is certainly in their best interest to do a good job as a registry. 2) updating the nameserver files is essentially trivial, and would be almost completely automated > It's also easy to imagine a situation where one of the registries feels > aggrieved about something and holds the whole process hostage when it's his > turn to do the domain merge. Any of the other sites could do it trivially. He would only hurt himself, because the names he registered wouldn't be included in the update. This is actually one of the strong points of decentralizing - -- *any* of the registries can do the whole job. > A jointly owned Swiss non-profit, like the > ones they've been talking about in newdom, with no axes of its own to grind > sure would be nice. It's the "no axes of its own" that is impossible to enforce, IMHO. [snip] > >In general, I would like to leave those > >relationships as unspecified as possible, to leave as much > >organizational freedom as possible. > > No argument there, except that I'm pushing everything down a level. I'd like > to define well specified protocols about how the first level customers talk > to a centralized light-weight registry for a shared TLD. Below that, > anything goes. > > >I guess another wrinkle is that a registry could serve several TLDs. > > I've been presuming that for the most part all customer registries will sign > up to serve every TLD they can, which is why it'd be nice to have a common > way for the customers to talk to the light-weight registries. > I hesitate to get into a discussion of light-weight registries, because you seem rather attached to the idea :-), but since you insist: I am uncomfortable with that approach because we are still left with a central point of failure, both human and technical, and we are still left with a potential monopoly point. Thus, light weight registries don't really solve the problem I would like to solve. We could talk about having the IANA require that LWRs be non-profit organizations, or put various other constraints, but those kind of restrictions I think are in the long run unworkable -- there are just too many different legal systems for the IANA to keep track of -- I have no idea what the legal definition of a non-profit corporation would be in Poland, for example. Even in the US laws vary on a state-by-state basis. Given that I don't say any realistic way for IANA to enforce organizational characteristics on a LWR, a LWR (or an affiliate) could then go into business as a "Full Service Registry", and compete unfairly against other registries servicing the same TLD. All my objections might be moot, if there were some significant technical difficulty that LWRs would allow us to avoid, but I don't really think there is. That is, I don't think the technical problems of doing Fully Distributed Registries are that great. So you can advocate Light Weight Registries, and I will advocate Fully Distributed Registries (Heavy Weight Registries seems so prejudicial :-)). However, I think it is important in the interests of progress that we not get to attached to these ideas, especially at this early stage. I would like to think that there are brilliant ideas lurking in the minds of the other people on this list that will make both of us think "why didn't I think of that?" - -- Kent Crispin kent@songbird.com "No reason to get excited", kc@llnl.gov the thief he kindly spoke... ---------------------------------------------------------------------- Date: 4 Aug 1996 12:01:43 -0700 From: johnl@iecc.com (John R Levine) Subject: Re: Lightweight vs. heavyweight registries >So you can advocate Light Weight Registries, and I will advocate Fully >Distributed Registries (Heavy Weight Registries seems so prejudicial >:-)). However, I think it is important in the interests of progress >that we not get to attached to these ideas, especially at this early >stage. I would like to think that there are brilliant ideas lurking >in the minds of the other people on this list that will make both of >us think "why didn't I think of that?" I guess we have different experiences dealing with ISPs. (My experience is too often described by metaphors regarding tying ones' own shoes.) Your concerns that a lightweight registry could be a single point of failure or a chokepoint are certainly reasonable, but I think there are non-technical ways to address them, primarily by making the centralized registry owned by as many of the next-level registries as possible. But I'm worried that in a fully distributed model, instead of a single point of failure, we'll have hundreds points of failure, each of which can temporarily screw up a shared domain. The problem is that running a shared domain requires two separate tasks: maintaining each individual registry, and merging the individual registries into the combined DNS database. Each registry needs to be good at the first, but there's a "tragedy of the commons" issue with the second. It seems likely to me that if shared domains are at all successful, there will be hundreds of registries sharing any particular domain. If there were, say, 365 registries and a domain was updated daily, this would mean that each registry would only do the merge once a year. This means that each night's update will be done by someone who at best did it once a year ago and more likely by someone who's never done it at all. Even if the merge process is straightforward, it is unlikely to be entirely mechanical and will need some human expertise to coddle it along. But for each registry, the worst they can do is to screw it up for one day each year, so it's not worth a lot of effort to get right. If you have 365 sets of data totalling 100,000 records, how likely is it that there won't be any inconsistencies or collissions at all? Assume that all data is 99.9% error free -- that's 100 errors. Without some sort of centralized scheme to enforce consistency and to resolve conflicts (e.g. when two registries each believe they registered FOO.UGH. first), I don't see much chance of ever getting a usable merged database. If someone has an idea for a fully distributed registry that addresses the consistency and update problems, I'd be happy to embrace it. But I haven't seen one yet. Geniuses, do your stuff. - -- John R. Levine, IECC, POB 640 Trumansburg NY 14886 +1 607 387 6869 johnl@iecc.com "Space aliens are stealing American jobs." - Stanford econ prof ---------------------------------------------------------------------- Date: 4 Aug 1996 15:06:46 -0700 From: Simon Higgs Subject: Re: WG charter timeline At 6:31 AM -0700 8/4/96, Kent Crispin wrote: >DRAFT > >Shared Top Level Domains Working Group (STLDWG) Charter > This can now be found at: I'll try and keep it up-to-date. ;) _____S_i_m_o_n___H_i_g_g_s_________________H_i_g_g_s___A_m_e_r_i_c_a_____ ... "I'm fine - it's the others" ............. President/CEO ............ _____e-mail: ___________ ______ .. ..