Minutes from Thursday Plenary discussion:
How about focusing on the end-user experience?
- Solve user-related out-of-the box zeroconf?
- How much do we have to design the upper layers in ensuring everything will run on IP?
- There are multiple sources of config info, hard to discern what/why something breaks
- What should we have put into SIP to make a VoIP phone easier to configure?
- The easiest parameter to configure is the one you don’t have to.
- Why not seek simplicity from the outset?
- Special cases for addressing, authentication, (other knobs) do not help
- Rather than “use-cases” …how about “user-cases” ?
- PCs are getting to the point where only my kids can configure them
- People who design the end-user experience are not here (at IETF)
- How about adding a new section to all future RFCs … on “User Experience”
- W3C had a team which focussed on schemas … produced a primer on how to do it (simple) …
- Enroll … we have a WG which not enough people are attending
Mandatory sections are evil
- Protocol Models:
- Very valuable new RFC4101
- How about putting some effort into disseminating the contents of that RFC to the community (e.g. at Sunday tutorials) ?
- Complexity dealt with in RFC3439
- Concern about the growing complexity and total number of protocols out there.
- IAB and IESG have been thinking/worrying about this for >10 years
- Lots of effort for configuring some devices (e.g. PCs) is to deal with non-standard complexities
- Should we invite a report from NANOG during IETF on “the TOP 10 things we are really upset about” ?
Layer Creep / Stack Creep:
- Concern that some functions are being reinvented in different layers
- We have TCP and UDP … why not just one?
- We promulgate multiple solutions (let many flowers bloom) … would it help to change this?
- Maybe we are spending too much time on layers 8-10
- Of course, there is lots going on at other layers which are beyond the IETF’s scope (i.e. domain of other standards bodies)
- (Protocol lipo suction ? Or full-blown surgery …)
- Infinite tunneling vs. shorter torso/waist height (remove vertebrae)
User experience and NATs
- How many of us help our relatives with their computers?
- … TOO MANY … we are failing !
- … or the scope of people’s expectations and universe is expanding … which complicates things
- We have no reasonable way for people to manage “walls” … our relatives need managed devices, enabled by reasonable signaling protocols
- Do we need an AETF ? (Applic’ns Eng’g TF) ?
- Subscription-based protocols ? Is the internet no-longer end-to-end?
- There are many sloppy implementations out there … maybe a lot of what we leave as user-configurable parameters should be hard coded
- Put some layers on a diet? The entire community would need to pull together to make this happen (i.e. recover some of the simplicity we used to have).
- Do we need it to use a single protocol at this stage of development?
- For IM or mail, we used a single protocol for a long time to transport store and forward messages …
- Features vs. complexity – hard to “add” simplicity
- - normal cycle is design, then simplify
- - imposing the above would run a risk of slowing our process(es)
- Is it still acceptable to publish standards-track RFCs that only address IPv4? We still have layer 3 work going on today which ignores IPv6
- OSPFv2 is IPv4-specific
- OSPFv3 is v6-specific … fact of life … sometimes we don’t have a choice
- “Acceptable” to whom?
- Experimental RFCs without text on v6 may be OK
Observation: perhaps more people would be stimulated to read more I-Ds if each came with a Powerpoint presentation (e.g. like an abstract / overview) ?
What if WGs had a page to present an overview of their technology? It is not a good assumption to hope everyone has read the drafts
- www.ietf.org and www.iab.org don’t appear to be running v6 on any server?
- Is this really the case? (Yes, except for the 3 weeks each year when IETF meets)
- What kind of example are we setting for the world ??
- Comment: NOC team did a great job to get v6 running over the Wireless network this week.
- Economic models … some influential stakeholders (e,g. network administrators) might want some consideration ….
- IETF structure: lots of WGs working on short term, fast time to market issues … and only 15 people (viz. IAB) looking at the bigger picture and the future
- 15 is not enough
- Compelling visions for how things will get better could stimulate different activities, attract different people, to the IETF … spawn a new ‘branch’ (or Area) within the IETF
- Aaron is trying to help the IRTF connect better with the IETF, and the IAB is also trying to encourage more of this
- We have lots of trade-offs to worry about:
- Balance takes time and consideration
- Designing simple protocols is really hard (like writing a message that fits on just 1 page?)
- Pain-shifting … always a lot of desire to shift the pain of complexity to somebody else
- Nobody is for reducing innovation … while innovation often generates more options (and therefore complexity) ?