Note: there are 2 plenaries to have minutes & presentations from - IESG and IAB. They need to be separate in the minutes. These are from the IESG. ---------------------------------------------------------------------- Scribe: Spencer Dawkins, with thanks to Jabber scribes * Harald Introductory Remarks IETF 57 attendance was down about 300 from normal, but attendees came from more countries than normal, too. Harald thanked local hosts, secretariat, Austria Telecom Future meetings will be IETF 58 - Minneapolis, November 9-14, hostless IETF 59 - Seoul, South Korea, February 28-March 5, hosted by Samsung, KIEF IETF 60 - probably back in USA Nearly 100 recent standards actions, including Diameter (AAA), MANET routing (exp), IPV6 addr arch revision, and IMAPv4 revision * David Black IPStorage Presentation iSCSI - encapsulation of SCSI on TCP/IP FCIP - encapsulation of Fiber Channel over TCP/IP, operating through gateways, using iSNS naming system Security - needed IPsec profile for IP storage Needed over half a dozen MIBs to manage all this Why IP Storage? Because we could - so many building blocks that could go into a SAN! GigE, switches, processors, IP, TCP, and economic reasons IP Storage almost done - total of 22 drafts, not meeting in Vienna, took about two years, with numerous implementations, especially of iSCSI MIBs developed in parallel with protocols and finished about the same time Engineering challenges were mostly in networking IPS worked with storage standards groups, even holding interim meetings with ANSI technical committees Organizational fun with distributed responsibility for components What worked well: - technical coordinators (copied from ISSLL WG) - dedicated efforts from primary draft authors - community effort to do "everything else" - design teams, with results opened to WG promptly (one design team completed its work in two months(!)) - multiple WG co-chairs (with "adult supervision" from Steve Bellovin) started out with three chairs(!) - "asking for help" - seems simple, but - MIBs, security, CRC, stringprep - Plugfest to catch problems early - Unwavering AD support (but didn't turn them into storage experts) - good relationships with other groups (IETF WGs across three areas, other standard groups) What Wasn't Perfect: - TCP framing tarpit - SRP patent mess - is confidentiality "MUST implement"? "Security ADs will get back to you" until NOMcom got back to former ADs - stretched mailing list (to be kind) - T11 meeting conflict with IETF in London - but can't believe there was only one conflict out of so many possible conflicts What made it fun: - spanned technologies and groups - IETF depth and breadth of networking expertise * Evolutionizing the IETF - status and plans - Harald "This is the most important part of my job at the IETF" POISED and its children had been going on for 10 years in 2001, and had turned into "nit-picking and noise" - started with IPR and NOMCOM to clarify existing procedures Dissatisfaction in 2002 was more obvious than the exact problem, but presence of problem was obvious We've been working for a year to get ready to work. Now we need to just do it Top of the tree is Goals, Standards Track, and Mgmt Avri - NOMCOM - making incremental changes to current process, no fundamental changes in any way. Has taken a year longer than expected. Making more detailed rules because of stresses to the trust model. Steve Bellovin - IPR - making incremental changes to current process. Has taken a year longer than expected. Have had a fundamental debate on re-chartering to change current policy and decided not to do so. Melinda - PROBLEM - chartered to produce problem statement and process plan, slightly behind schedule, but not a lot. Problem statement one rev away from last call. Some WGs using something like CVS for document control (although this is not widespread). Margaret - EDU - needed feedback on current efforts and direction for future effort. John - COACH - with the most amazing quotes... Margaret - IMPROVE - the proposal that was the road not taken... we need to talk Consistent theme is that traditional WG model does not seem appropriate for Goals, Standards Track, or MGMT! No deadlock getting started matters most of all... Harald - Summary David Perkins - understanding what consensus means - it's not a vote. And the Postel principle isn't on the list - interoperability is the goal. "macky" [missed name) - it would be an improvement if WG chairs were elected by WGs due to conflict of interests and ability to actually represent the WG. Scott Bradner - yes, yes, no, no, no, no, no. Have done an excellent job of teasing the problems out. "This process" is ambiguous? Not a WG, not a design team (don't know how to pick well - smartest under- graduates chosen by random lot). Need to solicit ideas and let them gel for a defined period. Fred Baker - appreciate PROBLEM group - identified a lot of problems, but not all of them. But one is staring us in the face. We don't know how to identify bellyachers and complainers here (think POISSON WG) and how to take them aside. Pete Resnick - "IESG can't hand down" - but we are? WGCHAIR workshops need to be published widely. Chairs are getting together to solve their problems. Is IESG doing this? Example of tracking system - what ELSE are you up to? Paul? Olaf Gunderson? Missed name - I've seen organizations implode on process, rules, regulations. We must avoid this. About a year ago, the IETF process was falling apart. Things are getting better. Problem helps us focus. Tracking helps. Nothing radical is necessary. Dave Crocker - this is the year after Yokohama - are things getting better? They seem to be... I thought not much had changed. Are things coming out more timely? Better quality? Are people happier? Ask yourselves - the answer matters. Suggestion - problems aren't other people's problems to fix - it's your responsibility to identify problems, write them up, and get backing for them. If you do that, no process matters. Erik Nordmark - overall, I think the problem statement document covers lots of things but overlooks technical issues that slow us down. We've solved all the easy problems. There's an installed base. We're raising the bar with security, privacy, etc. and there's a lot more to do. Can we build on top of something or do we have to write 200-page specs? Melinda - we've taken input from people who have provided it. We haven't had much input from people in certain roles, so they aren't in the document. Heshom Soliman - we should consider the variety of backgrounds in these process WGs. How does the IETF interface to the outside world? Culture clash, etc. and friction. There are also procedural issues - how do we bring a sample of the IETF into these WGs? Is there a uniform process that can help us speak the same language? Christian - I would start saying "no" earlier than Scott Bradner. "High Quality" is a problem for us - we need "sufficient quality". IETF wasn't doing standards 20 years ago - we were doing engineering. Now we are telling people what they should do. We used to tell people how to do things, not what to do. We used to use the market to help us decide. Our focus on high quality and power has paralyzed us. Alex Contras - you guys did such a good job you scared me. It looks like you know what you need to do to change the IETF, and that scares me because the IETF is special the way it is - can't change it in too radical a fashion. Must use small steps. Bernard Aboba - a lot of improvement can be done by the people in this room, if they want to make things better - think XML2RFC. It may be important to change the structure, but I'd set a deadline of 30 days to improve things with NO changes - a suggestion box that we could implement immediately. The perfect IETF is the enemy of the good. --- Daniel - Rough consensus and running code isn't working, and we're not focused on this idea year. Should replace our mantra with "running code and rough consensus"! Has someone actually tried any of the things we are bickering about? Tom Taylor - need to appreciate the change of environment on the IETF, and our success. Megaco had a positive result because of commercial conflict - we broke a deadlock. PWE3 had two camps. There are other examples. I'm sure there were holy wars in the past, but some thought should be given to these changes. Cut through deadlock? Accept deadlock? More people have read the problem draft than in many WGs... Core problems identified? 70-30 "has not identified" - come to the WG and help! Shared goal? 70-30 "shared goal". Process is helpful? 50-50. WGs? 20-80. Design team? 20-80. Pick someone to decide? 90-10. But *I* know what to do... John Klensin - you left off a question - should we make the IESG decide and explain their decision and be held responsible? 60-40. IAB? 20-80? What does "held responsible"? Alain Durand - in a company, the CEO makes the decision and is held responsible. Erik Nordman - but the IESG is overloaded... Scott Brim - you don't want a WG, you don't want an IESG. But you can do what looks like NOMCOM, and we know how to do that. Margaret - do we have to act? Can IETF go without organizational changes for three years? Thomas Narten - what is "organizational change"? Harald - we haven't had this in ten years. 40-60 that we can't. Bob Hinden - we're not going to die. "Survive" isn't the right word. There's also a SOLUTIONS mailing list. * IESG Open Mike Some changes from last time - Erik Nordmark is stepping down after years of service (applause), and will be replaced by Margaret Wasserman in the Internet Area (applause). Adam Roach - South Korea venue? Steve Bellovin - SK has highest penetration of broadband in the world. They care about the Internet. Avri - I could get invaded anywhere. (Kurt Lindquist?) - sometimes to see is to believe. Enjoy Korea! "DMZ Area" joke... (joke?) Ole Jacobson - this will be a great excuse to buy a new cellphone... James Kempf - they have CDMA-1 (don't know about roaming agreements) Jon Rosenberg - concerns about these facilities (distance, midnight subway closing, water, space in lobby, power) - is there a set of requirements written down? "Hosting criteria" updated? Need slides with larger fonts. I enjoyed walking - I walked off my food! And having no power wasn't all bad. Randy Bush - Net here has been excellent, especially wireless. Nice work. Rob Austein - we're insulting our local hosts, and that's probably over the top. Steve Bellovin - who needs to download presentations? A lot of people... --- power strips come up every meeting! Can we nail this down? Can they move the conference center closer to the city? No, but they apologized... NOC recognized in absentia. Raj Patel - it's not that all the US conferences were perfect! Best social in forever. --------- Daniel - look at the networking here, not just in this conference. Look at what we get for IP and the crap companies in the US deliver - flapping, one-way routing, security. We haven't fixed security. Have you guys been thinking about this? Will the Internet make TV look good? Randy Bush - building peer-to-peer protocols, so everyone can be a first-class citizen. Steve Bellovin - 80 percent of security problems aren't crypto problems, they are buggy code. Lixia - we're engineers, right? We know about open loops and about feedback. What feedback do we have about our output - protocols? Where is operational feedback? We have 134 WGs and 3 WG names that include the word "operations"? Say there are 30 - are there enough? Bill Summerfield - it would be useful to have more operational and end user feedback in all the WGs, not just the OPS area. Vendors have filtered requirements. Randy - IETF is an alliance of operators, vendors, and researchers. Operators had been trailing off, but some are coming back. But they don't get the response they are looking for. Headcount on distribution - we have a stronger alliance than we thought! [missed speaker] - need to think about working in all possible conditions. [missed speaker] - we're seeing more and more complex protocols that we can't provision. Randy Bush - as we make the space more complex, everything we do has to fit in with everything else and makes everything more difficult, especially if you're an operator - make stuff simpler and do only what we must do. Brian Carpenter - we had a lot of operators in the room, and they told us what they wanted (and then went away, but). There's a disconnect between operations participants and their management chain. Lucy Lynch - groups aren't transparent - no one says names or affiliations, a lot of complaints about things being deployed too early, and 10 percent of people read the drafts. If you aren't reading the stuff you're working on, you're going nowhere. Dave Crocker - we don't need to know affiliations, but - complexity hasn't made it into quality discussions, and that seems wrong. If software is more complicated, that means more code. Bob Hinden - Draft Standard give you implementation experience and lets you take stuff out. We should use this more. What can we do about spam? Not much - SPAM BoF decided we were ineffective. We do have ASRG in IRTF... but we won't solve spam here... Spencer Dawkins - where is the suggestion box for the organization? Harald - either a good idea or DoS bait - we'll figure this out on the Solutions mailing list.... ---------- End of minutes ---------- |