Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
-BEGIN PGP SIGNED MESSAGE- > "Harald" == Harald Tveit Alvestrand <[EMAIL PROTECTED]> writes: Harald> In the discussions leading up to this document, we actually had 3 Harald> different other levels of "inclusivity" up for consideration: okay, I very much like these descriptions. Harald> - "Everything that runs over the Internet is appropriate for IETF Harald> standardization". Harald> - "Everything that needs open, documented interoperability and Harald> runs over the Internet is appropriate for IETF Harald> standardization". Harald> - "Everything that builds infrastructures on the Internet that Harald> needs to be open and interoperable is appropriate for IETF Harald> standardization". These are three levels that I understand, and each seems to enclose the next. Harald> - "Everything that can seriously impact the Internet is Harald> appropriate for IETF standardization". This is very much more nebulous, because "seriously impact" is a question of very open judgement. ] Collecting stories about my dad: http://www.sandelman.ca/cjr/ | firewalls [ ] Michael Richardson, Sandelman Software Works, Ottawa, ON|net architect[ ] [EMAIL PROTECTED] http://www.sandelman.ottawa.on.ca/ |device driver[ ] panic("Just another Debian/notebook using, kernel hacking, security guy"); [ -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.2 (GNU/Linux) Comment: Finger me for keys - custom hacks make this fully PGP2 compat iQCVAwUBP5g84oqHRg3pndX9AQH01AP/ayMZ2WJVxz7xZXVSu9Pbew9U1A+GLUFb PVgK45qNL/qsL95U4cU1SyV5Tn2YYTjWkSD4j8tVNHAX+HyoqDJPYgWFwevOKblY HCwUj3N6Y/U43TpIZ8+w8NqcIkV0Z4BPc9kjpSjiUeTOZ4nfY+Pbg3yS+vaUvWcd ThqgtWgB7Lc= =p3P3 -END PGP SIGNATURE-
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> The number of application protocols with the oomph to "break" the > Internet is quite small OK, I've gotta ask - how many times do we break the Internet before we reverse this reasoning? (How many times is "too many"?) (signed) curious
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> The number of application protocols with the oomph to "break" the > Internet is quite small however, it's not safe to assume that it's zero. any new killer app that were poorly designed could do it. also, you might be underestimating the damage done by HTTP (1.0 or later).
RE: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
Christian, we might be looking through opposite ends of this tunnel. --On 16. oktober 2003 15:15 -0700 Christian Huitema <[EMAIL PROTECTED]> wrote: I think this point is one of the critical causes of conflict when talking about the IETF mission - and unless we lance the boil, actually talk about it, and attempt to *resolve* the issue, we will go on revisiting the issue forever, with nothing but wasted energy to show for it. Well, to paraphrase a well known leader, "the IETF, how many divisions?" The gist of this comment is that someone developing a network application protocol ought to somehow get a blessing from the IETF. Reality check. Who got the IETF approval to deploy ICQ, Kazaa, or for that matter HTTP? For application protocols, I view it in the opposite direction - if someone comes to the IETF and *asks* for the IETF's advice, blessing or ownership, what are the conditions under which we say "yes"? Or "no"? For those that never ask, and never become important, I say "not my problem". The number of application protocols with the oomph to "break" the Internet is quite small - offhand, I'd say that HTTP/1.0 probably was the closest try. If the Internet is so fragile that a poorly developed application can break it, then the IETF response should not be to try control each application. It has to be, design checks that can be implemented by cooperating hosts and routers so that their neck of the Internet is in good health! Now there's an idea. :-) The flipside is of course with those things that are *already* under IETF control, or critical for our infrastructure, for some reason. The abstracted version of the fights over MIME types, URI schemes, SIP extension etcetera seems to be "don't extend until you've talked to us about what you're doing, and if we don't like it, don't try to pretend that we did" (the P-headers, vnd. MIME types and the proposed faceted URI schemes); I'm not certain what the abstracted version of the fights over COPS, CR-LDP, RSVP-TE and so on are. The IETF has got fewer divisions than the Pope, of course. Anyone is free to ignore us. And we need to remember that, sometimes.
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
--On 16. oktober 2003 13:15 -0400 Eric Rosen <[EMAIL PROTECTED]> wrote: - "For the Internet" - only the stuff that is directly involved in making the Internet work is included in the IETF's scope. In other words, routing, DNS, and Internet operations/management. Adopting this as the IETF's mission would be a very radical change indeed! I erred in describing that category. I should have used something else - it was not what the IESG thought it was saying in its proposed mission statement, so me recycling the term "for the Internet" in the bulleted list I made added confusion rather than clarifying. Sorry! (I still think it's a valid point on the scale. It leaves us with a much smaller IETF. But some people tend to think that's a positive side.) Harald
RE: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> > According to you, this has nothing to do with the IETF. It might > result > > in the congestive collapse of the Internet, but who cares, the IETF > > doesn't do street lights. I would like to see the criteria which > > determine that telephones belong on the Internet but street lights > don't! > > thanks for making the most concise statement of the conflict here in the > discussion so far! > I think this point is one of the critical causes of conflict when talking > about the IETF mission - and unless we lance the boil, actually talk about > it, and attempt to *resolve* the issue, we will go on revisiting the issue > forever, with nothing but wasted energy to show for it. Well, to paraphrase a well known leader, "the IETF, how many divisions?" The gist of this comment is that someone developing a network application protocol ought to somehow get a blessing from the IETF. Reality check. Who got the IETF approval to deploy ICQ, Kazaa, or for that matter HTTP? If the Internet is so fragile that a poorly developed application can break it, then the IETF response should not be to try control each application. It has to be, design checks that can be implemented by cooperating hosts and routers so that their neck of the Internet is in good health! -- Christian Huitema
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
The example I'm thinking about involved predecessors to OpenGL. As this example doesn't even involve communication over a network, I would agree that it is out of scope. ... [OpenGL example] It's not that other examples such as X couldn't have used more network knowledge to avoid problems (e.g. the mouse stuff), but that the network stuff is the tail of that and many other dogs. Because of my employement history, I may know a little more about how to do graphics in general or over IP networks than many IETF participants, but I know that I'm abjectly completely utterly incompetent for doing exactly what the IETF started to do in that case. Great scope example. The issue for OpenGL, however, demonstrates a gap in as much as the developers would probably have liked something like dccp so that they could use a library to get Nagle, backoff, etc. While we're a wire protocol sort of a group, we all should realize the importance of generality and good library support ;-) If "out of scope" were removed as an acceptable reason to not do things, then you would never squelch bad efforts. An effort isn't bad because it's out of scope. An effort is bad because it's bad, and we invest our faith in the IESG that they will use good judgment to catch bad efforts. If anyone on the IESG does not feel empowered to say "no" they should not be on the IESG. WG chairs need to vet their own group's work first, of course. And we could certainly do a better job on that. Eliot
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
Scoping is certainly used successfully as an argument at the WG level, through the more common pronnouncement "that would require a change to the charter.." Scoping aids WGs in being able to move the ball forward in the direction of predfined goals, and hence is a process aid. This is scoping at a micro level. I would think that the role of mission is to provide scoping at a macro level, the kind of scoping that determines whether a WG is established in the first place or not. More importantly I would suggest, the simple requirement for making binary decisions about whether something is in scope or not is necessary but not sufficient. An institution surely needs some way to guide its priorities as well. So one could for example agree with Eric's definition of what the IETF's mission is, but once that is done, what then guides the priorities of the IETF? I think you will find this to be at the heart of the debate: scoping=>smaller workload=>focused differentiation in the standards marketplace+better quality output. Every entity must decide what it is going to do uniquly better than any other entity. This is the purpose of mission. Generic catchall missions do not help entities keep the eye on that particular ball.
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> From: Eric Rosen <[EMAIL PROTECTED]> > ... > > Sheesh!--next you'll be telling us that you never heard the phrase > > "out of scope" before last week. > > Sure I have. There's hardly a piece of work done by the IETF that someone > hasn't claimed to be out of scope. It's just that the phrase is not used > consistently. That's true. >If we look at the historical facts about the work that the > IETF has traditionally taken on, it's hard to draw any conclusion other than > that anything is in scope which promotes and facilitates the use of the > Internet and of IP infrastructure. And I think that's exactly what the IETF > should be doing. That's wrong. At best it's meaningless. For example it supports lobbying Congress. > > The example I'm thinking about involved predecessors to OpenGL. > > As this example doesn't even involve communication over a network, I would > agree that it is out of scope. ... It was out of scope, but not because it did not involve putting graphics stuff over UDP or TCP, because it did. My fellow employees in SGI's network group and I breathed a sigh of releaf when Ron returned from an IETF meeting to report that "out of scope" had carried the day against other IETF participants who thought that knowing people who knew about Nagle and congestion control and avoidance was enough to design graphics remote procedure calls or similar. It's not that other examples such as X couldn't have used more network knowledge to avoid problems (e.g. the mouse stuff), but that the network stuff is the tail of that and many other dogs. Because of my employement history, I may know a little more about how to do graphics in general or over IP networks than many IETF participants, but I know that I'm abjectly completely utterly incompetent for doing exactly what the IETF started to do in that case. > > Often the brutal WG chairs say they don't think the WG knows enough, but > > it's the scope arguments that carry the day. > > I've never had much luck myself with scope arguments, unless they could be > backed up with an argument either that the center of expertise is elsewhere, > or that the topic has no bearing on IP. Of course, people will sometimes be > willing to agree that the center of expertise is elsewhere without > necessarily agreeing that they themselves aren't experts ;-) Sometimes scope > arguments are merely face-saving ways of saying "we don't know what we are > doing". Other times, scope arguments are merely "polite" ways of saying "we > don't think you know what you are doing". You almost never hear someone > saying "that sounds like a really good idea, but unfortunately it is out of > scope". Yes, with the proviso that you mean you usually don't hear people really meaning that last sentence. You certainly hear those words a lot. If "out of scope" were removed as an acceptable reason to not do things, then you would never squelch bad efforts. I suspect the whole effort of defining IETF charters or missions is a very bad idea. It's often better to not spell things out, but to rely on the good judgement of the people running the show. Vernon Schryver[EMAIL PROTECTED]
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> The gist of this comment is that someone developing a network > application protocol ought to somehow get a blessing from the IETF. > Reality check. Who got the IETF approval to deploy ICQ, Kazaa, or for > that matter HTTP? The fact that someone did something without the IETF's approval does not imply that what they did is outside the scope of the IETF, or that it is beyond the IETF's mission.
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> Sheesh!--next you'll be telling us that you never heard the phrase > "out of scope" before last week. Sure I have. There's hardly a piece of work done by the IETF that someone hasn't claimed to be out of scope. It's just that the phrase is not used consistently. If we look at the historical facts about the work that the IETF has traditionally taken on, it's hard to draw any conclusion other than that anything is in scope which promotes and facilitates the use of the Internet and of IP infrastructure. And I think that's exactly what the IETF should be doing. > The example I'm thinking about involved predecessors to OpenGL. As this example doesn't even involve communication over a network, I would agree that it is out of scope. But that's a rather extreme case, most of the contentious areas do involve communications over an IP infrastructure. > Often the brutal WG chairs say they don't think the WG knows enough, but > it's the scope arguments that carry the day. I've never had much luck myself with scope arguments, unless they could be backed up with an argument either that the center of expertise is elsewhere, or that the topic has no bearing on IP. Of course, people will sometimes be willing to agree that the center of expertise is elsewhere without necessarily agreeing that they themselves aren't experts ;-) Sometimes scope arguments are merely face-saving ways of saying "we don't know what we are doing". Other times, scope arguments are merely "polite" ways of saying "we don't think you know what you are doing". You almost never hear someone saying "that sounds like a really good idea, but unfortunately it is out of scope".
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> From: Eric Rosen <[EMAIL PROTECTED]> > > That is wrong or at least a gross overstatement. > > If that's what you think, I invite you to make a list of all the > IETF-standardized protocols and explain how they are all (or even more than > 50% of them) needed to make the Internet work. There's a progression here: 1. ad hoc network interoperability group forms. 2. it has some success and gains some fame. 3. it is besieged by people eager to borrow its printing press 4. it is besieged by people who know everything about everything and have a duty to write or at least control all standards on everything including what people perversions do in the privacy of their own networks. I've been complaining about #3 for many years. Examples of #4 include some of the more vigorous combatants in the IPv6 site local arena and the notion that notion that nothing is out of scope. > > There have been many things that the IETF has chosen to step away from but > > that ran and run over the Internet. Some graphics standards come > > immediately to my mind ... Those graphics standards were kept out of the > > IETF not because the working groups involved thought they didn't think > > they were experts, but because the subject was out of scope for the IETF." > > I'm not familiar with this particular case, but I don't see why protocols > for distributing graphics would be thought to fall outside the scope of the > IETF, any more than protocols for distributing voice or video. Of course, > graphics standards that have nothing do with distribution of the graphics > over IP would be out of scope. The example I'm thinking about involved predecessors to OpenGL. People who know about network stuff know enough to stuff bits into wires, but that's the earier part of things like OpenGL, Microsoft's alternative whose name eludes me, JPEG, MPEG, and so forth. > > No committee is ever able to limit itself on grounds of insufficient > > expertise. > > Now, there is a gross overstatement! For everyone who proclaims himself > (rightly or wrongly) to be an expert on some topic, there are always two > other people who claim that he is clueless. The other two base their claims on their own greater expertise and wouldn't dream of suggesting that they are not well suited for standardizing whatever it is. > It's not uncommon for a WG to > refuse to pick up a topic because the consensus is that the topic's > proponents are clueless. Please name an example of such a case. I have seen WG chairs and others use brute force and out-of-scope arguments to halt nonsense, but I've never seen "we don't know enough" work. Often the brutal WG chairs say they don't think the WG knows enough, but it's the scope arguments that carry the day. Sheesh!--next you'll be telling us that you never heard the phrase "out of scope" before last week. Vernon Schryver[EMAIL PROTECTED]
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> That is wrong or at least a gross overstatement. If that's what you think, I invite you to make a list of all the IETF-standardized protocols and explain how they are all (or even more than 50% of them) needed to make the Internet work. > There have been many things that the IETF has chosen to step away from but > that ran and run over the Internet. Some graphics standards come > immediately to my mind ... Those graphics standards were kept out of the > IETF not because the working groups involved thought they didn't think > they were experts, but because the subject was out of scope for the IETF." I'm not familiar with this particular case, but I don't see why protocols for distributing graphics would be thought to fall outside the scope of the IETF, any more than protocols for distributing voice or video. Of course, graphics standards that have nothing do with distribution of the graphics over IP would be out of scope. > No committee is ever able to limit itself on grounds of insufficient > expertise. Now, there is a gross overstatement! For everyone who proclaims himself (rightly or wrongly) to be an expert on some topic, there are always two other people who claim that he is clueless. It's not uncommon for a WG to refuse to pick up a topic because the consensus is that the topic's proponents are clueless.
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> From: Eric Rosen <[EMAIL PROTECTED]> > > - "For the Internet" - only the stuff that is directly involved in making > > the Internet work is included in the IETF's scope. > > In other words, routing, DNS, and Internet operations/management. Adopting > this as the IETF's mission would be a very radical change indeed! While > this particular mission statement does seem to reflect the interests of a > certain notorious IESG member, let's not pretend that this has ever been the > limit of the IETF's mission. The IETF has always been concerned with things > that make the Internet more useful, and with things that expand the utility > of the IP protocol suite. There's never been a time when "for the Internet" > was an accurate representation of the IETF's concerns. That is wrong or at least a gross overstatement. > ... > The formulation I like is "Everything that needs open, documented > interoperability and runs over the Internet is appropriate for IETF > standardization". This is much truer to the IETF's current and historical > practice. That is also wrong or at least a gross overstatement. There have been many things that the IETF has chosen to step away from but that ran and run over the Internet. Some graphics standards come immediately to my mind. > That doesn't necessarily mean that the IETF has to standardize everything > that falls within its mission. For instance, a particular area might fall > within the mission, but the IETF might not have the expertise to tackle it. > A WG in that area could then be rejected on the grounds of "insufficient > expertise". Such decisions would have to be made on a case-by-case basis. > Again, this is the way such decisions have always been made in the IETF. No committee is ever able to limit itself on grounds of insufficient expertise. The people who join committees are predisposed to think that they're sufficiently expert to deal with the subject matter. Those graphics standards were kept out of the IETF not because the working groups involved thought they didn't think they were experts, but because the subject was out of scope for the IETF. That most of participants had no clues about computer graphics was incomprehensible to most of the participants, and unfortunately irrelevant. Vernon Schryver[EMAIL PROTECTED]
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
> - "For the Internet" - only the stuff that is directly involved in making > the Internet work is included in the IETF's scope. In other words, routing, DNS, and Internet operations/management. Adopting this as the IETF's mission would be a very radical change indeed! While this particular mission statement does seem to reflect the interests of a certain notorious IESG member, let's not pretend that this has ever been the limit of the IETF's mission. The IETF has always been concerned with things that make the Internet more useful, and with things that expand the utility of the IP protocol suite. There's never been a time when "for the Internet" was an accurate representation of the IETF's concerns. You are of course welcome to propose such a radical change to the IETF's mission. But if you are going to circulate a document under the subject line "IESG proposed statement on the IETF mission", you should make it clear that the IESG is proposing to make a complete change in the IETF mission. Instead, you give the impression that the IESG thinks that "for the Internet" is and has always been the IETF's mission. The formulation I like is "Everything that needs open, documented interoperability and runs over the Internet is appropriate for IETF standardization". This is much truer to the IETF's current and historical practice. That doesn't necessarily mean that the IETF has to standardize everything that falls within its mission. For instance, a particular area might fall within the mission, but the IETF might not have the expertise to tackle it. A WG in that area could then be rejected on the grounds of "insufficient expertise". Such decisions would have to be made on a case-by-case basis. Again, this is the way such decisions have always been made in the IETF.
Re: IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
% --On 15. oktober 2003 12:57 -0400 Eric Rosen <[EMAIL PROTECTED]> wrote: % % > Well, let's test this assertion. Suppose a consortium of electric % > companies develops a UDP-based protocol for monitoring and controlling % > street lights. It turns out that this protocol generates an unbounded % > amount of traffic (say, proportional to the square of the number of % > street lights in the world), has no congestion control, and no % > security, but is expected to run over the Internet. % > % > According to you, this has nothing to do with the IETF. It might result % > in the congestive collapse of the Internet, but who cares, the IETF % > doesn't do street lights. I would like to see the criteria which % > determine that telephones belong on the Internet but street lights don't! % % thanks for making the most concise statement of the conflict here in the % discussion so far! % I think this point is one of the critical causes of conflict when talking % about the IETF mission - and unless we lance the boil, actually talk about % it, and attempt to *resolve* the issue, we will go on revisiting the issue % forever, with nothing but wasted energy to show for it. % % In the discussions leading up to this document, we actually had 3 different % other levels of "inclusivity" up for consideration: % % - "Everything that runs over the Internet is appropriate for IETF % % - "Everything that needs open, documented interoperability and runs over % the Internet is appropriate for IETF % % - "Everything that builds infrastructures on the Internet that needs to be % open and interoperable is appropriate for IETF standardization". % % - "Everything that can seriously impact the Internet is appropriate for % IETF standardization". % - "For the Internet" - only the stuff that is directly involved in making % the Internet work is included in the IETF's scope. % % a discussion argue based on "the mission of the IETF", with conflicting % definitions, is not the best thing for the Internet. % % Harald I guess for me, I always thought that the IETF and its precursors were interested in developing engineering solutions / designing protocols that would allow "end2end or any2any" communications, regardless of underlying transport media, be it seismic wave, avian carrier, radio waves or the PSTN. - At no time did I ever truly beleive that the systems that used these protocols/solutions would always be on and fully connected. Infrastructures that use IETF products have nearly always been only partially connected and many systems are not always on. So while a design goal might have been to support always on/fully connected state, the reality is that infrastructres have nearly always been disjoint/unconnected and endpoints come and go. But when they are connectable, they should function in a seamless, e2e fashion, at least IMHO. And then you neglect an unstated presumption in the last two bullet points: As perceived by who? --bill Opinions expressed may not even be mine by the time you read them, and certainly don't reflect those of any other entity (legal or otherwise).
IETF mission boundaries (Re: IESG proposed statement on the IETF mission )
Eric, --On 15. oktober 2003 12:57 -0400 Eric Rosen <[EMAIL PROTECTED]> wrote: Well, let's test this assertion. Suppose a consortium of electric companies develops a UDP-based protocol for monitoring and controlling street lights. It turns out that this protocol generates an unbounded amount of traffic (say, proportional to the square of the number of street lights in the world), has no congestion control, and no security, but is expected to run over the Internet. According to you, this has nothing to do with the IETF. It might result in the congestive collapse of the Internet, but who cares, the IETF doesn't do street lights. I would like to see the criteria which determine that telephones belong on the Internet but street lights don't! thanks for making the most concise statement of the conflict here in the discussion so far! I think this point is one of the critical causes of conflict when talking about the IETF mission - and unless we lance the boil, actually talk about it, and attempt to *resolve* the issue, we will go on revisiting the issue forever, with nothing but wasted energy to show for it. In the discussions leading up to this document, we actually had 3 different other levels of "inclusivity" up for consideration: - "Everything that runs over the Internet is appropriate for IETF standardization". Obviously, that might cause some reactions from organizations like the W3C, OMG, ISO, ITU, the power grid standardizers, the bank transaction standardizers and others even if the IETF were able to gather the required competence, it's hard to see how we could build a management structure that could handle "everything". - "Everything that needs open, documented interoperability and runs over the Internet is appropriate for IETF standardization". A bit smaller, but still huge, and hard to draw boundaries around. Advantage: Everything we currently work on is unquestionably part of the IETF's scope. - "Everything that builds infrastructures on the Internet that needs to be open and interoperable is appropriate for IETF standardization". This would place SMTP, DNS and LDAP (in the original vision) inside the IETF's sphere, but would leave the traffic lights (and the current way LDAP is used) outside it. - "Everything that can seriously impact the Internet is appropriate for IETF standardization". Argues for keeping HTTP and DNS, would include your hypothetical traffic lights, but would probably leave POP/IMAP out, and leaves people arguing about both SIP and L3VPN. - "For the Internet" - only the stuff that is directly involved in making the Internet work is included in the IETF's scope. It's far from clear in my mind what the right thing is, or what the appropriate path forward is if the IETF regards its purpose as being one or the other - we might, for instance, decide that we standardize stuff that needs to be open and interoperable, but have different evaluation criteria for those things than for those things that "make the Internet work", and will dispose our resources accordingly - I don't know. And if we decide that certain things we currently do are outside our scope, we've got a responsibility to make sure the work effort is handled in a responsible fashion. But it's relatively clear to my mind that continuing to have both sides of a discussion argue based on "the mission of the IETF", with conflicting definitions, is not the best thing for the Internet. So - rather than stating something completely vague, we put out a proposal. If it's the wrong proposal, it should be changed. But please be specific about what you think it should be changed to. makes sense? Harald