Internet-Draft Dinners and Lunches November 2020
Richardson Expires 1 June 2021 [Page]
shmoo Working Group
Intended Status:
Best Current Practice
M. Richardson
Sandelman Software Works

How often and how long should IETF virtual meetings be?


This document recommends a model for IETF virtual meetings that emphasizes new work, community building and cross-area concerns.

This document recommends virtual meetings be planned a reduced amount of overlap among sessions, with short sessions with generous gaps.

Status of This Memo

This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."

This Internet-Draft will expire on 1 June 2021.

Table of Contents

1. Introduction

IETF107 (Vancouver) was the first pandemic IETF virtual meeting held in March 2020. IETF108 (Madrid) was the second pandemic IETF virtual meeting held in July 2020. IETF109 (Bangkok) was the third pandemic IETF virtual meeting held in November 2020.

IETF110 (Prague), will be held as a virtual meeting in March 2021. It is unknown at the time of this writing whether IETF111 (San Francisco) will be possible.

1.1. IETF 107 experience

IETF107 consisted of a light week with no more than about 4-hours of sessions over Monday to Thursday. There were multiple tracks simultaneously, but not more three sessions at any one time. Priority was given to newly formed WGs that had not yet met, to Birds of a Feather session, and to cross-community sessions including specifically "dispatch" sessions.

These sessions were done with XXXX (webex? right?)

The timezone was approximately "Vancouver"-ish. Most sessions started in late morning, which turned into mid-evening in Europe (going into midnight), and middle of the night in Asia.

All other WGs were assigned a day in the weeks following IETF107 from March 30 to April 30. The WGs were asked to pick a time (and time zone) convenient to them, and they scheduled webex meetings on "", although a few WG chairs decided to use "Zoom". Many meetings occured in the North-America/Europe optimal time slot of 1500UTC ("10am EDT").

Despite the "pick a time", a small number of WGs managed to pick times when other WGs were meeting, and there were in fact collisions.

IETF 107 would be best described as a marathon rather than a sprint.

1.2. IETF 108 experience

IETF108 consisted of a very heavy week.

Eight simultaneous tracks were scheduled, although with slightly shorter meeting times of 50 minutes and 100 minutes. The plenary occured at the normal Wednesday afternoon time, and other constants like "saag" being on Thurdsay afternoon were also maintained.

The day started in late-morning "Madrid" (Central Enropean) time and ended before dinner time. The longest break was a 30 minute break after the first session. Other breaks were in the 10 minute range.

This accomodated East-coast North-Americans, but Pacific coast participants had to get up for 4am. The schedule accomodated those in Asia slightly better, but it was still middle of the night in New-Zealand for all but the first sessions.

All WG sessions used meetecho with additional features having been added to support things like humming, and some slightly better queue management.

Many participants experienced a typical number of conflicts. See, for instance

Some WGs declined to schedule meetings at IETF108. This was due to a combination of scheduling, timezones and the fee to attend. WGs that did this usually had a healthy and regular set of virtual interim meetings that were ongoing.

To some, IETF 108 felt like a sprint: others felt that the pace, while intense was okay if one paced oneself well.

1.3. IETF 109 experience

IETF109 consisted of a heavy week similiar to IETF108.

Eight simultaneous tracks were scheduled. Each day had a two hour slot, a thirty minute break, a one hour slot, another 30 minute break, and then a two hour slot.

Survey feedback said that participants perferred that the meeting time be as compressed as possible, so the total meeting duration was 6 hours long.

The day started in afternoon "Bangkok" (UTC+7hours: 1pm) time and ended before dinner time.

Participants on the West coast of North America (California, Vancouver, Seattle, etc.) were 16 hours behind, so the meeting time of 0600 UTC worked out to start at 2100 (PST) the night before, and went until 0300 (PST). It was a late night for those participants.

Participants on the East coast of North America (Toronto, Boston, NYC, Atlanta, etc.) were 13 hours behind, so the meeting time of 0600 UTC worked out to start at 0000 (EST) the night before, and went until 0600 (EST). Participants either slept during the day (7am to 3pm) and then got up and spent the evening with their family, or they slept during the afternoon/evening (3pm to 10pm), and got up and joined "first thing". Those who slept during the day usually did so because that's when their house was quiet (partner at work, children at school): they did the meeting at the end of their day. Those who slept in the evening, seemed to cite being able to collaborate with European colleagues after the meeting, during the European "day"

Participants in Europe found the meeting start time of 0600UTC meant that they had get up a bit early. (This experience was similiar to the IETF108 experience for East Coast north Americans) They meeting then occupied most of their day.

At the time of this writing, no feedback from Pacific/Asian participants was received. Input is sought.

Side meetings were a challenge: a few had side meetings during the 30 minute breaks, but many people socialized on This was also used for one on one follow-up discussions. Getting more food, drink, and attending to washroom needs were also important to most participants.

A few side meetings were called for after the meeting time, that is after 1200UTC. This was convenient for Europeans and Asian participants, and East Coast participants who slept in the evening, but it did not work for West Coast participants at all.

IETF109 seemed much better attended than IETF108.

2. Suggestions for future meetings

2.1. Principles for for Recommendations

There is great utility in having all of the IETF members present and available together. Doing this in-person has many great benefits (and many costs internal and external), not enumerated here.

An unconstrained meeting with few conflicts and a few empty slots in each individual schedule maximizes the benefit of the togetherness. Some call it Working-Group "tourism", others call it "impromptu cross-area review", but there are significant benefits to having "bored" participants with time on their hands wander into a WG to which they have little knowledge.

To be fair, there is some significant negative experience with mic comments like, "I haven't read the document, but..."

Yes, often these comments do lead to significant amounts of friction. Time wasted at the mic explaining that that were covered in the document.

Sometimes, though, they lead to an understanding that two WGs are facing the same issues, and that they should work together. And getting this understanding is really quite valuable.

This is PARTICULARLY the case for new work (BOFs), and for the first few meetings of a newly formed WG. One of the important thing for many individuals to internalize what the boundaries of the new work is, and what the overlap with the work they are doing is. This often comes about from the discussions that go around slide ware, to seeing who else cares about this work, and why.

Heavily conflicted (physical) meeting schedules have been a growing issue, with the result that many people do not have the time (due to conflicts), or the energy (due to exhaustion) to pay attention to the new work.

2.2. Recommendations on meeting structure

2.2.1. Encourage Virtual Interim meetings for ongoing work

This document therefore recommends that the IETF encourage a healthy growth of virtual interim meetings which are:

  1. heavily focused.
  2. frequent participant-time-zone optimized
  3. typically occur at twice-monthly to monthly intervals appropriate for getting work done.

There is a significant self-selection problem with virtual interims and participants.

In groups where the use of virtual interim meetings are used a lot, the choice of time zone in which to hold the virtual interim meeting can determine who attends. When the virtual interim time zone selects against participants from a particular part of the world then they tend to participant less. If the working group then does a poll of participants as to what time zone to use, then having lost participants from the less popular time zone, the result is likely to reinforce this selection.

This suggests that virtual interim meeting time zones should not always be chosen according to popularity (such as "doodle" poll). But, at the same time, a working group which is highly focused and doing good work should be forced to work at inconvenient times. There is a definite tussle here!

The IESG is encouraged to find ways to insert some variation in meeting time. This needs to be done with care, and with some amount of sharing of the pain among different working groups as well. 1500UTC

The 1500 Central European time (in whichever Daylight Savings is active), corresponds to 7AM Pacific, 10AM Eastern, 2PM (UK), 3PM (Paris), 4PM (Helsinki), and 10PM Beijing. This time period is used for IESG Thursday meetings, and is popular because it accomodating to many participants. It is hostile to Eastern Australia, New Zealand and Hawaii participants. Fundamentally, the lack of people who live in the Pacific ocean biases the time zones to those which are daytime in Europe, which is essentially in the middle.

It is therefore difficult to "share the pain" -- those at the extremes (Pacific North America, and Asia/Pacific) will find the times picked to almost always be annoying. To be clear: while a meeting could be planned for early evening (16:00) in San Jose, which is early morning in Tokyo (9AM), it would be outside of office hours, or even night for most everyone else in the world.

2.2.2. Allow IETF Virtual meeting week to focus on community-wide activities

While the IETF mission statement is very outcome focused, that doesn't mean that every activity needs to be only outcome focused. I think that the IETF "plenary" meeting times should be arranged to not just permit, but encourage cross-working-group participation.

This document suggests that the IETF virtual meeting week be focused on:

  1. governance and meta-activites like IESG, IAB, NOMCOM
  2. *DISPATCH activities
  3. BOFs for new work
  4. growing the virtual Hackathon
  5. newly formed WG meetings,
  6. maybe WGs who have gone through some major milestone, and are for instance, rechartering

Rather than attempt to compress the schedule down so that the time between the beginning of the day and the end of the day is as short as possible, we should instead arrange the schedule with generous break intervals to accomodate adhoc 1:1 or rather-small-group meetings.

That is, the "hallway session" needs some time and space to be useful.

In particular, having some gaps where a person can reasonably expect to track down someone else to chat about a specific thing, or just to catch up.

We used to stimulate the "hallway" discussion using cookies and beverages.

Often this is the best way to talk though DISCUSSes among ADs and document authors.

Those that used the system, found it was exceptionally useful for this.

2.2.3. Number of Tracks for virtual meetings

This document recommends that we have fewer tracks. Somewhere between two and four.

That is, like IETF108 and more like IETF107. A few more tracks, and slightly longer days.

This document does not recommend that the schedule be adjusted to accomodate all the time zones. Because of where the Pacific Ocean is, that pretty much always puts noon somewhere near Greenwich.

The 1-1-1-* process should mean that the locale chosen for the cancelled physical meeting defines the time zone. Participants should be expected to "travel" to that time zone: many physical trips take 24 to 36 hours of airplane time, and asking participants to take a similiar amount of time to adjust their body clocks is not unreasonable.

Once a year, the participants will find they have to do almost no adjustment, when the meeting is "near" them.

Meetings should be restricted to between a 4 and 7 day period though, which gives them some time to adjust before "returning" to work the next week.

Historical meetings have accomodated in excess of 200 hours of session time. (Eight hours/day X 4.5 days X eight tracks = 288 hours, minus some for the plenary).

The virtual meeting should restrict itself to approximately 100 hours of session time. 6 sessions of 50 minutes, with ~30 minute breaks between, over 4 tracks, and over 4 days gives 90 hours of session time. This calculation does not count time for a plenary where there is only one track. This formula is intended to be a guideline only.

Note that for many, conflicts mean that they have to view the session after the fact via Youtube. While the "play back at 2x speed" can be effective for some listeners, this can result in significant number of extra hours of time for the meeting.

Rather than compete for available plenary-week WGs slots, WGs should compete to demonstrate that they don't need a slot.

Having a plenary-week slot should mean that additional "supervision" is required :-) Either because the WG is very new, or because the WG is old and has lost it's way.

3. Security Considerations

The excessive use of caffeine and sugar to adjust jet lag may have health effects on the participants. In addition, there is some possibility that decisions made under such influence might negatively affect the security of the Internet. Multiple reviews after periods of "sober second thought" should catch any such poor decisions.

4. IANA Considerations

This document makes no requests to IANA.

5. Acknowledgements


6. Changelog

7. Normative References

Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, , <>.

Author's Address

Michael Richardson
Sandelman Software Works