Experiences from the first fully-online IAB workshop on Network Impacts of COVID-19

While it’s not new anymore that things have to move into the online world, as the COVID-19 Network Impacts Workshop 2020 report has just been published as RFC 9075, it seems like a good time report back on this first-time experience for an IAB workshop.

The challenge was that IAB workshops are meant to be highly interactive and “locking” a well-composed, small group of people into the same room for a day was a main part of the event concept. Even though that specific part was not possible, we can surely say that the workshop was quite a success, and also identified new opportunities that are worth pursuing in future.

The topic of the impacts of the COVID-19 crisis on the network naturally came up earlier that same year and the IAB decided to cover this topic by a workshop in order to get a better view on what happened behind the scenes: what went well, and what went potentially wrong or needs more effort to improve in future. Long story short: we can say that the general understanding at the workshop was that the Internet survived this significant change in traffic volume and usage well, however, there are many places where we probably need more data and it might also not be fully clear if we were “just lucky” to some extent. More information about the technical discussion can be found in the recently published workshop report.

But back to the general workshop experience first. For the IAB it was clear that this topic needs timely discussion and therefore an online workshop was the only option. Initially the IAB left the concrete organization of the workshop open and decided to first collect input while only indicating a selected week as the planned time horizon. After review of the contributed papers, the proposed topics could be nicely clustered into 2-3 buckets with a well-covered set of papers that described measurements during the first lockdown in 2020 from various different angels.

Based on this input, it was clear that the workshop should start off with a couple of short presentations on this data, followed by one or two sessions that would allow for more open discussion as is usually done in IAB workshops. Given the well-known challenges with time zones and the fortunate opportunity that we could spread out these sessions over a full week (rather than squeezing everything into one day to minimize travel effort), we started a poll to find the most suitable two-hour time slot for the participants, and asked people to reserved the respective slots for Monday, Wednesday, and Friday.

Speaking of participants, so far IAB workshops have been invitation-only, mainly because room size for these kinds of events are usually limited to roughly 30 people. Given this limit doesn’t apply to virtual events, of course the question came up whether the workshop should still be invitation-only. However, the room size is not the only reason for this approach. 

Beside the physical limits of room size, the interactive discussion needed for a successful IAB workshop works better with a “reasonably” sized group. And this is still to some extent true for online meetings. Of course it doesn’t hurt necessarily if more people listen quietly but the more people who participate without contributing, the less often others are incentivized to speak up. In a smaller group of invited people, it is more clear that the expectation is also that those people are supposed to actively contribute. But, we recorded and published all sessions for those that only wanted to listen. 

There is another practical reason: we needed input from the position papers to organize the agenda of the workshop. Requiring a position paper to get invited of course provides an incentive to actually write such a paper, and also ensures that those people writing a paper are interested enough in the topic to invest that (limited amount of) time. In the end, we invited everyone who submitted a paper, even though some papers were broader in scope than we intended for the workshop. However, we were interested in having those people also involved in the discussion.

Splitting up the workshop into three sessions over a whole week turned out to be a great choice. Of course the selected time slot wasn’t perfect for all participants, but it seems people were interested enough to stay up late for these limited number of slots. A real positive effect was that we always had one day off in between the sessions. While we didn’t realize this opportunity when planning, it turned out to be a real benefit in order to have some mail discussion or dynamically adapt and prepare for the next session based on the discussion of the previous one.

As we all know by now, there are also drawbacks to this setup. The social part was more or less entirely missing. As in many cases, we did benefit from the fact that a lot of participants already knew each other. We got a bunch of contributions from academia, but even though those people might not show up to (all) IETF meetings anymore, many of them were to some extent known faces. Still, even when you at least know most people already, the part where everybody sits together in the evening and keeps reflecting about discussions from the workshop was entirely missing and is probably usually also an important factor for a successful IAB workshop. 

So while the workshop was  an overall success, and the opportunity to reflect and adapt between different sessions is something worth preserving (maybe with an online pre-meeting?), we also look forward to meeting in person again because those informal interactions, outside the scheduled presentations, are an important part of how new ideas arise.

Stay healthy and hopefully see you soon!