Minutes of the 2014 IAB Retreat, 8-9 May 2014, Cancun, Mexico
- Jari Arkko (IETF Chair)
- Mary Barnes
- Marc Blanchet (via WebEx for items 1-6, then in person)
- Alissa Cooper (IESG Liaison)
- Lars Eggert (IRTF Chair)
- Mat Ford (ISOC Liaison)
- Joel Halpern
- Ted Hardie
- Joe Hildebrand
- Russ Housley (IAB Chair)
- Eliot Lear
- Xing Li
- Cindy Morgan (IAB Executive Administrative Manager)
- Erik Nordmark
- Andrew Sullivan
- Dave Thaler
- Brian Trammell
- Kathy Brown
- Sally Wentworth (via WebEx, present only for item 3)
- How can the IAB Lead the Community?
- The A of the IAB
- Internet Governance
- Improving IAB Programs and Workshops
- Program Goals for 2014
- RFC Series
- Toronto Plenary
- I* CEO Meetings
- How the IAB Interacts with Other Bodies
- Guidance to ISOC
- Root Servers
- Encryption in the Open Internet
- Liaison Oversight
- Process vs. Technical
- Electric Power
- Retreat Evaluation
1. How can the IAB Lead the Community?
Russ Housley asked the question, “How can the IAB lead the community?” and opened up the floor for brainstorming.
Eliot Lear noted that his goals for the IAB in 2014 include surviving the ongoing Internet governance debates, re-invigorating the IAB Programs, and work on UDP at layer 3.5. Brian Trammell agreed that he would like to see technical recommendations out of the UDP 3.5 work.
Ted Hardie stated that he would like to see the IAB do more work on privacy and security, in conserving the value of the network. He suggested that the IAB should be more radical in the technical work it encourages the community to develop, and used UDP 3.5 as an example of that. He advised the need for the IAB to define a system architecture for the Internet as a whole to keep all of the pieces together. Overall, he said, the IAB should do work, and entice others to do work as well.
Xing Li noted that the Snowden Effect has increased the risk of a split Internet, citing China’s firewall as an example of such a split Internet. He added the China has a larger percentage of Internet users than any other country, but that China’s technical community is still quite small. He suggested the need for a unified Internet community modeled on the same sort of rough consensus the IETF uses. Xing also noted that the transition from IPv4 to IPv6 brings risks of splitting the Internet, and that the IAB should make efforts to ease that transition and help those who are working to build the future Internet.
Joel Halpern said that, given the IAB’s responsibility for external relations, the IAB should work with the open source community to communicate their value to us, and our value to them. Mary Barnes added that we should look at the work within the IAB Programs on open source and see that what we are delivering is actually useful.
Dave Thaler noted that originally, the Internet architecture was designed to allow connectivity between anyone and anyone else, but today’s users care more about preventing unsolicited inbound traffic, while preserving their own outbound access to outside content. He suggested that there is work to be done to fix the current Internet architecture and reevaluate the best ways to meet the needs of today’s users.
Jari Arkko asked how much is achievable, and what are the first steps? He suggested that the IAB has authority to speak, and that articulating some of the current challenges may be the first step.
2. The A of the IAB
The IAB charter [RFC 2850] states:
The IAB provides oversight of the architecture for the protocols and procedures used by the the Internet.
Dave Thaler led a discussion of the IAB’s role in architectural oversight, noting that if there are several different ways of doing something, ideally one method should be chosen in order to avoid duplication of the same protocol functionality. In reality, there are multiple competing protocols from the IETF, other SDOs and consortia, and the vendor community, and a trend towards reinventing things at a higher layer to go with each organizations existing lower stack.
During the Technical Plenary at IETF 70 (and in RFC 5218), the IAB presented a list of factors that contribute to a successful protocol:
- Positive Net Value (Meet a Real Need)
- Incremental Deployability
- Open Code Availability
- Freedom from Usage Restrictions
- Open Specification Availability
Dave Thaler noted that many of the success stories were from protocols that originated outside of the IETF, and where technical quality was not a primary factor in their success. A common pattern in the IETF is that when there are multiple proposals, the conclusion reached is that the requirements are such that one solution does not fit all. In those cases, either two or more independent solutions are standardized, or else common building blocks are established to create new solutions. This often leads to users not having sufficient information to choose intelligently between the options.
Dave Thaler added that a variation of the option where multiple competing solutions are developed is that neither solution is standardized (e.g., only Experimental RFCs), and it is left for the market to choose.
Eliot Lear suggested that an additional option would be to reject the conclusion that one size does not fit all, and further constrain the requirements in order to develop something smaller that would fit in a single protocol.
Brian Trammell noted that how the IAB provides guidance will depend on what the conflict is (e.g., competing protocols with total overlap in their design goals versus competing protocols with significant but not total overlap in their design goals).
Mary Barnes noted that the reasons for choosing one solution over another are often driven by business interests, rather than technical reasons. Ted Hardie agreed, adding that when different communities have different goals, it is difficult to resolve what the correct technical outcome should be.
The IAB agreed that its role is to help the Internet community choose intelligently when there is more than one solution available, by providing guidance for when each solution is appropriate, when the IETF does not appear to be able do so itself. Joe Hildebrand suggested asking the IESG for input on areas that could use IAB guidance.
3. Internet Governance
The IAB discussed the ongoing Internet governance activities, including ICANN 49, the World Telecommunication Development Conference (WTDC), the World Summit on the Information Society (WSIS), and NETmundial. Eliot Lear stated that the IAB’s role is to provide the technical expertise needed to run the Internet. He noted that not all parties involved in these discussions believe in the multistakeholder model. All of the discussions thus far will feed into the discussions at the ITU Plenipotentiary Conference (Plenipot) later this year.
Sally Wentworth will be leading the ISOC delegation at Plenipot; she will begin an offline discussion about whether someone from the IAB should be involved in that delegation.
Russ Housley asked if ISOC would examine the contributions for Plenipot and provide a summary of issues that have technical implications. Sally Wentworth replied that ISOC would, and that the proposals seen to date have primarily been at a regional level. Russ thanked Sally and asked her to make that matrix available to the IAB once it is ready so that the IAB can provide input.
Sally Wentworth noted that some countries are looking for ways to expand the mandate of the ITU. One way the IAB can address that is to articulate clearly why the existing mandate continues to be relevant. The IAB agreed to take further discussion of Plenipot into the ITU-T Coordination Program, where they will help Sally come up with that messaging.
The IAB also discussed the Internet governance implications of the ongoing discussions about the potential transition of IANA out of NTIA. The IAB discussed whether these discussions should happen in IAB IGOVUPDATE sessions at IETF, or if there should be a General Area Working Group on the topic. Andrew Sullivan noted that the split will depend on which parts will require IETF consensus before proceeding. He suggested that the IANA Strategy Program go through the list of existing issues and separate out what should be discussed where.
Eliot Lear asked if there are other resources needed in order for this to be successful. Andrew Sullivan replied that neither the IETF nor the IAB would be the ones negotiating the agreement with NTIA. Russ Housley added that we would need an RFC to document any update to the MOU between the IAB, ICANN, and IANA.
The IAB briefly discussed the frustrations in China regarding TISec. Xing Li reported that the group working on wireless security in the IEEE standards space is also working on modifications to IPsec, but are looking at 3-way handshakes instead of 2-way handshakes. The TISec is not a Chinese National Standard at this moment. Russ Housley noted that the IESG declined to assign a protocol number for TISec, and asked how TISec works as a national standard without the assigned number. Xing found later that TISec currently uses number 253 (use for experimentation and testing defined by RFC3692).
5. Improving IAB Programs and Workshops
Ted Hardie noted that the IAB Program structure was instituted to help find points of leverage with which to apply force, as well as to help retain institutional knowledge as IAB membership shifts over time. He proposed that as the IAB has some level of independence, the IAB also has a responsibility to suggest more radical solutions. In order for that to work, Program members must be sufficiently independent (though not from IAB oversight) so as to make unpopular decisions.
Ted Hardie suggested that in order to increase the IAB’s level of influence, Programs must include a certain amount of outreach and attract individuals who can affect the outcomes; successful Programs would pull in energy from throughout the entire Internet technical community. He added that while being radical was not a prerequisite for a Program, the Program must still be capable of being radical.
6. Program Goals for 2014
The IAB reviewed the goals and current work items for the IAB Programs.
6.1. RFC Editor Program
Joel Halpern noted that the RFC Series Oversight Committee (RSOC), aka the RFC Editor Program, was mandated by RFC 6635. The Program performs the review of the RFC Series Editor (RSE), ensures the RSE’s priorities, and provides additional review of the RSE’s results.
6.2. Liaison Oversight Program
Eliot Lear reported that the operational goal of the Liaison Oversight Program is to facilitate review of the IETF’s liaison relationships. The Program is working on simplifying and consolidating the RFCs relating to liaison management, as well as talking about potential new activities and relationships. He noted that over the past year, the Program has not done much work to recruit new liaison managers, as the IAB has been sending public calls for volunteers. Dave Thaler said that it would be helpful if the Liaison Oversight Program would work on recruitment again, in order to help with the current efforts to find a liaison manager for ISO/IEC JTC1/SC22.
Eliot Lear reported that Robert Sparks and the IETF Tools Team have been working on improvements to the Liaison Tool.
6.3. ITU-T Coordination Program
Joel Halpern noted that the ITU-T Coordination Program’s recent efforts have been focused on addressing potential overlaps between IETF and ITU-T work. With the upcoming ITU Plenipotentiary Conference scheduled for later this year, Joel expects the Program to have additional work items around responding to Plenipot proposals.
6.4. IP Evolution Program
Dave Thaler reported that current work in the IP Evolution Program has focused on architecture, with several documents in process (e.g., draft-iab-host-firewalls, draft-iab-2870bis, draft-iab-filtering-considerations). There has been difficulty getting consensus on what work should be done. Sometimes this is normal because the IAB wants to tackle difficult issues.
Dave Thaler reported that there have been discussions about splitting the IP Evolution Program into two Programs, with one focusing on TCP/IP Evolution (evolution of the transport later, UDP 3.5, NAT translation and tunneling, TAPS) and one focusing on DNS evolution and naming more generally, including resolving names via different protocols (DNS, mDNS, host files, etc.), gTLDs, split horizon, DBOUND, DNS encryption, etc. Others observed that it should be named more generally than just “DNS” evolution.
The IAB agreed to split the program into two Programs, add the UDP 3.5 work to the TCP/IP Evolution part of the Program, and to possibly rename the Programs to something more accurate in the future.
6.5. Internationalization Program
Andrew Sullivan reported that the Internationalization Program’s only work item has been an update to RFC 2277 (IETF Policy on Character Sets and Languages), but that the work has currently stalled. Much of the Program’s energy has been devoted to advising the RFC Series Editor on internationalization as the RFC format evolves.
After discussion, the IAB agreed to move the Internationalization Program into a dormant state (no active work items), while remaining open to advise the IAB on internationalization issues as they arise. The Program will also continue to work on the update to RFC 2277 as time permits.
6.6. IANA Strategy Program
Andrew Sullivan reported that the IANA Strategy Program will spend much of the next year focusing on the proposed transition of the IANA function away from NTIA.
6.7. Emergency Services Program
Mary Barnes reported that the Emergency Services Program has seen little progress on their deliverables over the past year, and suggested that the Program be moved to a dormant or concluded state.
Russ Housley noted that the IETF has done a lot of work on ECRIT but has run into problems because the European Union cannot reference the RFCs, and so they have asked ETSI to develop a solution for emergency services. Alissa Cooper asked how the IAB could evangelize the IETF solution to the rest of the world. Andrew Sullivan suggested that the IAB could work on a document outside of the current Program structure. Brian Trammell added that there is concrete technical guidance on this topic that the IAB could provide, in conjunction with political guidance from ISOC.
Alissa Cooper took an action to see if the ECRIT Working Group Chairs and RAI ADs have bandwidth over the next six months to contribute to a document explaining why emergency services are better served over ECRIT.
6.8. Privacy and Security Program
The IAB agreed to merge the existing Privacy Program and Security Program into one Privacy and Security Program with the following purpose statement:
The IAB Privacy and Security Program is a successor to its previous Security and Privacy programs. It provides a forum to develop, synthesize and promote security and privacy guidance within the Internet technical standards community. While security and privacy have been each been explicitly and implicitly considered during the design of Internet protocols, there are three major challenges which face the community:
- Most Internet protocols are developed as building blocks which may be used in a variety of situations. This means that the security and privacy protections each protocol provides are also necessarily piecemeal and that default requirements for use are commonly either missing or ill-understood.
- Many security approaches have presumed that attackers have resources on par with those available to those who secure the system. Pervasive monitoring, distributed networks of compromised machines, and the availability of cloud compute each challenge those assumptions.
- Many systems breach the confidentiality of individuals’ communication or request more than the minimally appropriate data from that communication in order to simplify the delivery of services or meet other requirements. When other design considerations contend with privacy considerations, privacy has historically lost.
This program seeks to consolidate, generalize, and expand understanding of Internet-scale system design considerations for privacy and security; to raise broad awareness of the changing threat models and their impact on the properties of Internet protocols; and to champion the value of privacy to users of the Internet and, through that value, as a contributor to the network effect for the Internet.
Ted Hardie noted that this was a significant amount of work for the Program, and suggested structuring the Program such that there a number of appointed focus area leads to drive work on particular topics.
The IAB agreed with this structure and set of work items. Ted Hardie will draft a call for volunteers for the new Program. The members of the existing Privacy Program and Security Program will be informed of the changes before anything is made public.
6.9. Tools and Processes Program
The IAB agreed to close the Tools and Processes Program, with existing work items being handled as action items on an as-needed basis.
The IAB discussed the potential transition of IANA away from NTIA. Andrew Sullivan suggested that at this point, the IAB should wait until ICANN makes their proposal before doing anything further, but agreed that the IAB should continue to monitor the situation closely.
Eliot Lear noted that the cultural differences between the various stakeholders might prove challenging when trying to participate in the discussions. Alissa Cooper added that the IAB should help ensure everyone has an opportunity to participate. Eliot asked how the IAB should engage with those communities.
Xing Li stated that he believes the IANA transition is difficult for developing countries to understand. He suggested the need for a simple English language description of the issues at hand, as well as an FAQ. Xing volunteered to start work on such a document.
8. RFC Series
Joel Halpern reported that the RFC Series Editor’s current priorities include:
- format update work
- updating the RFC Editor website
- reviewing the errata system
- digital object identifier (DOI) assignment
- creating a digital preservation policy
- author assistance activities
9. Toronto Plenary
The IAB discussed the Technical Plenary for IETF 90. Andrew Sullivan circulated an updated proposal for the names and DANE topic. The IAB brainstormed potential speakers and ways to further refine the proposal.
Joe Hildebrand agreed to work with Andrew Sullivan on an updated version of the proposal, including potential speakers.
10. I* CEO Meetings
Russ Housley noted that the I* meetings were originally intended as a way for the leaders of the various organizations in the Internet ecosystem to get together and communicate informally so that they did not make announcements that surprised each other. IANA has been a topic of discussion since the first meeting.
The Montevideo meeting in October 2013 provided a timely opportunity for the group to make a statement in light of the then-newly released Snowden revelations on mass surveillance. Since then, there has been pressure to issue a statement coming out of each of the I* meetings. Jari Arkko noted that it is important for the IAB and IETF to set up their positions ahead of time so that nothing surprising comes out of those statements.
Andrew Sullivan agreed that while the I* Leaders’ meetings are important in that they allow the different bodies to coordinate with each other and avoid surprises, he is skeptical of the push towards issuing statements, as it is not always clear that the various bodies agree what those statements mean. Eliot Lear agreed with that sentiment.
Overall, the IAB agreed that it was important to prepare for these meetings, so that those attending can be confident that they are expressing and supporting the agreed-upon opinions of the IAB.
The next I* CEO meeting will be held in Istanbul in September, in conjunction with IGF.
The IAB approved publication of draft-iab-host-firewalls as an Informational RFC. Russ Housley will forward the document to the RFC Editor on behalf of the IAB.
12. How the IAB Interacts with other Bodies
The IAB discussed how it interacts with other bodies (e.g., SDOs, governments) in cases where there is not a liaison relationship in place. Eliot Lear noted that Thomas Narten has been working in the Liaison Oversight Program to draft text on this topic.
Andrew Sullivan suggested the need for a set of principles by which to organize these communications, so that people can talk of behalf of the IAB without the IAB needing to come to formal agreement about what to say for each and every case.
13. Guidance to ISOC
Earlier in the retreat, Kathy Brown asked what outcome of the potential transition of IANA outside of NTIA would be considered unacceptable. The IAB agreed to provide guidance to ISOC on this topic, noting, among other things:
- the desire for one global Internet
- the risk of destroying the Internet commons
- the risk of having multiple DNS roots
- the risk of splitting the mechanism by which names and IP addresses are allocated
Eliot Lear agreed to begin drafting this guidance to ISOC.
14. Root Servers
Marc Blanchet reported that the RSSAC is still in the process of reorganizing internally. The RSSAC and the IAB have been working on two documents to update to RFC 2870; the RSSAC is focusing on the protocols, and the IAB is working on the deployment considerations. draft-iab-2870bis will progress as a Best Current Practice document that is AD-sponsored in the IETF stream.
15. Encryption in the Open Internet
The IAB discussed encryption in the open Internet, and whether there should be a move away from cleartext into more encrypted data. Xing Li noted that in China, with the Great Firewall, encrypted data is likely to be blocked; cleartext is easier to verify, debug, and troubleshoot. He asked whether the move could be towards encrypting the metadata only.
Ted Hardie noted that cleartext traffic across the network was not put into place for the protection of the users. Any variety of bad actors can read that traffic, including traffic for financial transactions. For that reason, he argued, the IAB should continue to foster the ability to use encryption. Ted noted that the question of encrypted metadata was discussed during the STRINT workshop, and that in many cases encrypting the metadata only protects the bad actors.
Joel Halpern disagreed; he said that it is not the network’s business to care whether data is encrypted or not, but that it is the IAB’s concern because the IAB worries about end-to-end traffic. He added that having a country cut itself off entirely from the network because the traffic is encrypted is not a reasonable outcome.
Erik Nordmark noted that intrusion detection systems are easier when traffic is not encrypted. He added the governments or companies may not care if traffic is encrypted within their borders, but that they do care when that traffic crosses their borders.
Marc Blanchet added that during the STRINT workshop, the group discussed the need for a document that describes the implications of a fully encrypted network. There are operational consequences, and those are architectural considerations that must be kept in mind.
Dave Thaler attempted to summarize the conversation thus far:
- An open Internet should allow users to connect to anyone
- in some cases that requires cleartext
- in other cases that requires encryption
The open question is which mode should be primary. Xing Li suggested having the ability to fall back to cleartext in cases where encryption fails. Dave Thaler replied that the decision to fall back to cleartext should be under the end users’ control, and they they should be able to opt out of any defaults.
Ted Hardie stated that the work for the Privacy and Security Program would be to write a document that makes a statement about how to handle this problem when designing new protocols for the Internet, and give examples of the implications. The IAB does not want to see an end result where users withdraw from the network due to lack of trust, or where the contributions of entire countries are lost from the whole of the Internet.
16. Liaison Oversight
Eliot Lear asked IAB members to consider joining the Liaison Oversight Program. He noted that there are new relationships in progress with ECMA TC39 and ISO TC-204; those relationships will eventually need IAB Liaison Shepherds. The role of the Liaison Shepherd is to check in with the Liaison Manager and make sure that any relevant information gets reported back to the IAB as a whole.
17. Process vs. Technical
The IAB discussed the current balance between process/political work and technical work. Several people voiced opinions that the technical work is being drowned out by the politics, however, they noted that it is important for the IAB to participate in the current IANA and Internet governance discussions.
The IAB considered splitting future retreats into one day devoted to technical issues and one day devoted to political issues, but concluded that some of the topics are too intertwined for that to be feasible.
Dave Thaler noted that while there are IAB Programs devoted to technical work and IAB Programs devoted to political work, it is the more political topics that tend to be raised back to the full IAB. Ted Hardie suggested that IAB members interested in certain topics should join the Program mailing list and be involved in the discussions earlier, rather that nit-picking at the end of the process. Eliot Lear added that the board should use Programs to resolve the issues that they can, and any issues (political or technical) that cannot be resolved should be raised to the full IAB.
18. Electric Power
The IAB discussed power consumption for mobile networks and mobile devices. Ted Hardie suggested that the IAB may want to revisit this topic for a tech chat or technical plenary sometime in 2015. Dave Thaler noted that power and cost was listed as one of the work items for the IP Evolution Program, but that the work was deferred in favor of other topics because there was interest by others in tackling the work inside the IETF. Dave explained that Suresh Krishnan has started to drive work on this topic within the IETF.
19. Retreat Evaluation
The IAB reviewed the structure of the retreat. The board agreed that the joint day with the IESG was useful, and recommended continuing that in the future.
The IAB agreed that co-locating with LACNIC was valuable, noting that several IAB and IESG members were able to participate in LACNIC sessions. Russ Housley suggested looking to co-locate with APNIC or AFRINIC for future retreats.