IETF7 Technical Plenary Thursday December 6 2007 Vancouver, Canada Notes by Mirjam Kuehne 1. Welcome 2. IRTF Chair's report (Aaron Falk) 3. IAB Chair's report (Olaf Kolkman) 4. What Makes For a Successful Protocol? (Dave Thaler) draft-iab-protocol-success-01.txt 5. Energy Engineering for Protocols and Networks (Bruce Nordman, Elwyn Davies) 6. Open microphone session ------------------- 5. Energy Engineering for Protocols and Networks (Bruce Nordman, Elwyn Davies) Elwyn Davies: Why might the IETF care? Is there a way when we design protocols to keep the amount of totally used energy down? It would be nice if we can keep the impact of global energy use down. Pleased to see the Low Power BoF earlier this week. Introduces Bruce Nordman who gives the rest of the presentation. 6. Open microphone session - Comments related to Energy Presentation - Henning Schulzrinne: How can we motivate IPv6? The answer was partially given today (IPv6 without NATs) IPv6 as the green protocol? :-) Joel Jaeggli: interesting. But most of the observations are basic facts of life of the semi conductor industry: increase of densitity decrease power. We approached these problems before and solved them. 16 kW per rack isn't that bad. Doesnt' see this as a dramatic issues. Looking at the mobility industry might be more interesting. Elwyn: would be good if we wouldn't make the problem worse by asking for more silicon in the core. Danald Eastlake: it is dictated by peak requirements. This will impact the IETF and we will have to come up with optical routing protocols to accommodate this. Elwyn: pure optical routing is nice and will hopefully be a feature for the future, but not ready today yet. Difficult problem. Iljitsch van Beijnum: If you send an ethernet packet to a switch or router you have to wake up the router. People were pushing to make these packets larger. Smart vendors broke the IEEE standards by inventing bigger packets. But it has to be configured manually. He has a draft to remove the necessity for manual configuration. Jeffrey Hutzelman (?): IETF has two WGs dealing with these issues: 6lowpan and hopefully rl2n (Routing For Low Power and Lossy Networks) - about tiny battery devices - Comments related to Dave Thaler's presentation - Bob Briscoe: related to protocol design success factors: One of the factors mentioned in the presentation noted that protocols are more successful if they don't disrupt existing business models. He would like to warn to use this success factor strictly. If we would have applied that, there wouldn't be IP. Dave Thaler: Quite possible that changing business model can be a good thing, but you have to convince them of that. Keith Moore: excellent piece of work. People might say though: "The IAB says NATs are successful" There are other concepts than success that we should not ignore. Dave: specific text suggestions? Dave Oran: Agrees, we should not set up a framework based on a Darwinwean standpoint. Ralph Droms: Are any of those successful protocols younger than 12 - 15 years? We need to be paying attention to what the IAB is saying here, so that the newer protocols are also successful. Dave Thaler: Success often doesn't become clear before sufficient time has passed. Loa: Right, you will not have protocols that are wildly successful that are only 1 year old. Bernard Aboba: The IETF can look out there in the open at successful protocols and see how they can be improved. Aaron Falk: if we make changes, we can think about those things and increase deployment. Leslie: technical superiority is not necessarily a factor for being successful. Older protocols brought into the IETF might also be more successful, because new work was more elastic and more experimental in nature. It was easier to bring in new ideas. Nowadasy we have to check if things are successful before we start looking at them in order to not bring the whole industry down. This is a challenge. Tony Hain: Would the IETF have approved BitTorrent? Or if so, would we have killed it? We should not always subsume things that work on their own. Bernard: if it is wildly successful, the IETF could probably not bring it down. Jonathan Rosenberg: great document and work. Founds the document actually more related to the IESG. Most sucecssful protocols are those that would not been approved by the IESG? What does that mean? Keith: Jonathan must have misunderstood the document. Success as defined by this document is merely ONE criterion. Yes, we should look at those criteria. IESG does not only look at success factors when approving a protocol. Leslie: WGs should also consider this work. Spencer Dawkins: good document, helpful for the IESG. Would it be reasonable for the IAB to look at some of the current protocols and measure against these criteria? Olaf: has the feeling that one of the more important concepts is already actively brought into discussion this week. Has heard of economical alignment and deployability being criteria in WGs. Eric Rescorla: There are some hard truth about the world. The IETF is not an organisation that wants to kill protocols. But there are WGs that should be killed. We should do this much more often. Henning: Can't we reflect back on protocols that were killed, but did survive anyhow? One example of that was brought up last night. But there might be others. In the WGs we are trying too hard to predict what is going to be successful. And this is happening in the the WGs, not the IESG. On the application layer we have not done anything new and interesting since mid 90s that has had wide scale impact. This seems to have moved out of the IETF. Dave Oran: to put a more positive spin on applications scenario: the tool kit for applications is much richer than it used to be. We might be putting ourselves out of business for a good reason. Loa: We sometimes looked at protocols who were developed outside and decided not to take them on, some of them became successes. We should look at those cases and see why they became successful and why we decided not to take them on. Eliott Lear: do you believe that the IESG has sufficient visibility in all this to make good decisions? Dave Oran: doesn't know. If you look at the instances those questions are most applicable, most of them are not for the IESG, but for participants, WGs etc. The only one that can be put into that category is the IESG's decision about BoFs. Wes Hardaker: one that is not discussed in the document and that may be a reason why protocols are outside outside the IETF is that protocols have a high turn over rate. It is about adopting protocols into the IETF. Bob Hinden: in hindsight it is easy to tell what is successful. It is much more difficult when you are designing it. Sometimes you just cannot tell. It maybe successful later for other reasons. Spencer Dawkins: It would be good if the IAB would have visibility into successful factors when new work gets chartered. What kind of questions can people ask themselves before they bring new work into the IETF? Keith Moore: for a lot of people the frustrating things is that they spend years on protocol design and then have the protocol not be successful. If you took your sample for the case study, only things that started 10 years ago were evaluated. Did they meet the success criteria as we defined them now or did they fail and if so why? Olaf: gave DNSSEC some thought in that context. This has been in the IETF for a long time. One success criteria: Is there perceived benefit? Hard to sell for a security mechanism. There is very little the IETF can do except by making the case why things are important. The market place decides. Mark Crispin: It is simply impossible to satisfy 2 or more masters at the same time. There are always various demands on a protocol. One often gets one of the masters angry or in an ideal case the agreement is just 'good enough' but nobody is really happy with it. The processes of the past can not be applied today. Are there other organisations that have a faster turn around and the same diversity as we have? An entity with only one vision and only one master to be satisfied is naturally faster. But we should not try that. Our diversity is our strength. Douglas Otis: It is good to worry about power. The network not repsonsible for most of the consumption. Where can we make a bigger difference? How can we clean up the mess (e.g. all the criminal activities that is being used by this technology). Bruce Nordman: Don't concentrate on 'what if's'. Instead look at what can we do today. What is worth doing today? And go do it. And good data matters. Berry Leiba: if we can safe energy for instance by using video conference instead of travel, we should look into that. The fact that the Internet has safed energy doens't mean we should disregard areas where we can improve and the energy we can safe on the Internet (???) Erik Rescorla: If the Internet is only using half percent of all the power, then we should not worry about that and instead look at other places. |