- From: Mark Nottingham <mnot@mnot.net>
- Date: Fri, 11 Jul 2014 19:30:32 +1000
- To: Willy Tarreau <w@1wt.eu>
- Cc: Jeff Pinner <jpinner@twitter.com>, HTTP Working Group <ietf-http-wg@w3.org>
On 11 Jul 2014, at 7:26 pm, Willy Tarreau <w@1wt.eu> wrote: > Maybe I missed something, but the underlying issues that led to the proposal > to remove CONTINUATION were complexity, processing cost and sensitivity to > DoS. So while the HOL blocking and buffering are probably not impacted by > the presence or not of CONTINUATION, these other issues definitely are. Or > am I off-topic ? There are lots of pathological ways to increase processing cost — e.g., 1-byte DATA frames. We’re not talking about limiting those (and I will be EXTREMELY unsympathetic to anyone who brings it up after this). Complexity has many facets, as we’ve heard; it’s not a clean-cut decision. AIUI the DoS scenario had more to do with HOL blocking and cost of teardown; if there’s a DoS vector that’s specific to this scenario, let’s hear it. Thanks, -- Mark Nottingham http://www.mnot.net/
Received on Friday, 11 July 2014 09:31:00 UTC