Skip to content

2015 Report

Mark Nottingham edited this page Mar 8, 2016 · 10 revisions

HTTP Workshop 2015 Report


In July, nearly 40 engineers got together for the first HTTP Workshop in Münster, Germany. Over four days, we discussed the use and evolution of the Web's protocol. These people included not only those active in the HTTP Working Group, but also engineers that don't necessarily have the time to track such efforts closely.

The agenda was loosely structured around a few "anchor" presentations and lots of time for discussion and hallway meetings. These included:

Besides the discussion "in the room", there were also a number of hallway discussions and social events, including a welcome drink by Akamai, and dinners hosted by Mozilla and Google.

HTTP Workshop 2015 Attendees

We also had a number of impromptu presentations, on topics like HTTP load balancer implementation concerns, a high-performance HTTP/1 parser, HTTP/2 prioritization and HTTP/2 testing.

On the last day, we brainstormed a list of ideas for future consideration in HTTP. It's long and very tentative, but interesting reading none the less.

See also Daniel Sternberg's blog entries: 1, 2, 3, 4.

Feedback and Future Workshops

As the Workshop closed, we sent out a poll to participants to gather feedback, and see if it was a good idea to hold similar events in the future.

The responses made a number of things clear:

  • People felt that the Workshop was valuable for them to attend.
  • All of the content got favourable feedback, but the most positive response was to the generous amount of unstructured time for discussion.
  • Respondents felt that holding an event like this about once a year for about three days would be good.

The feedback included many positive comments, such as:

Face to face interaction is useful. Brainstorming future projects together in a room is so much more effective than doing so on a list.


I loved the way the workshop was organized. The sessions had lots of discussions (and discussions spur new ideas more than presentations of what-we-already-understand things). And it seemed to me that because the session was organized as such that the hallway discussions became more interesting as well. Even though we all work with HTTP very deeply, not all of us share our methodologies and experiences, especially using public channels. Exchanging the different views and experiences (due to the difference in how we work with the protocol) could not have been more valuable.


As someone new to both the broader discussion of HTTP and to the group of people involved in this area, I found the workshop helpful in both getting to know others in this field as well as becoming better aware of the current issues and future directions. In particular, I found some of the discussions around the intertwining of the framing and semantic layers of HTTP/2 extremely insightful, as it helped better set the stage for understanding QUIC and other potential plans for improving the transport layer for HTTP/3.

We also got a number of suggestions for improvements to the format, etc., as well as who we invite:

I heard a statistic mentioned that ~50% of the attendees of this HTTP Workshop were new to this community (i.e., they don't attend the IETF HTTP WG meetings, etc.) -- I am one of these "new" people. I think that's great! That said, I think the workshop could be improved if it encouraged/solicited input from more attendees -- oftentimes, the discussions tended to be focused on a back-and-forth between a couple of people, with everyone else silently listening.

A few people noted in their feedback what a number of us had noticed and discussed: we need more diversity in this community, especially women.

Overall, it's pretty clear that holding an event like this is valuable, because it gives a rare opportunity for the HTTP community to come together and reflect upon how the protocol is used, discover new perspectives and understandings, and explore how it might change in the future.


The attendees of the 2015 HTTP Workshop were (in alphabetical order):

  • Andreas Garkuscha (Apple)
  • Arnaud Braud (Orange)
  • Barry Leiba (IESG, Huawei)
  • Brad Fitzpatrick (Google)
  • Dan Druta (AT&T)
  • Daniel Stenberg (Firefox, Curl)
  • Eric Rescorla (Mozilla)
  • Erik Nygren (Akamai)
  • Gaetano Carlucci (Politecnico di Bari)
  • Georg Koppen (Tor)
  • Herve Ruellan (Canon)
  • Hooman Beheshti (Fastly)
  • Ian Swett (Google)
  • Igor Sysoev (NGINX)
  • Ilya Grigorik (Google)
  • Jana Iyengar (Google)
  • John Border (Hughes)
  • Julian Reschke (Program Committee, Greenbytes)
  • Kaoru Maeda (Lepidum)
  • Kazuho Oku (DeNA, H20)
  • Marcus Ihlar (Ericsson)
  • Mark Nottingham (Program Committee, Akamai)
  • Martin Thomson (Program Committee, Mozilla)
  • Mike Bishop (Microsoft)
  • Mike Perry (Tor)
  • Mike West (Google)
  • Moritz Steiner (Akamai)
  • Moto Ishizawa (Yahoo! Japan, Sasazka)
  • Patrick McManus (Mozilla)
  • Poul-Henning Kamp (Varnish)
  • Rob Trace (Microsoft)
  • Roy Fielding (Program Committee, Adobe)
  • Salvatore Loreto (Ericsson)
  • Scott Marshall (Apple)
  • Stefan Eissing (Greenbytes)
  • Valentin Bartenev (NGINX)
  • Willy Tarreau (HAProxy)

HTTP Workshop 2015 Attendees

You can’t perform that action at this time.