- From: Eliot Lear <lear@cisco.com>
- Date: Mon, 25 Nov 2013 08:30:53 +0100
- To: Mike Belshe <mike@belshe.com>, James M Snell <jasnell@gmail.com>
- CC: Tim Bray <tbray@textuality.com>, Alexandre Anzala-Yamajako <anzalaya@gmail.com>, Matthew Kerwin <matthew@kerwin.net.au>, Michael Sweet <msweet@apple.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
Hi, On 11/24/13 8:22 AM, Mike Belshe wrote: > But I don't want to hold up progress on HTTP/2.0 for everyone because > of a very minority claim that legacy hardware performance is bad. You're arguing a false premise. That is- web sites that are looking at aggregate costs in a competitive market compare against performance of their peers. This factors out legacy h/w. I hesitate to go further because the data here can lead to ambiguous interpretations, but it's a sure bet that there will be classes of environments that have certain performance characteristics that make HTTP/2 palatable, and others where perhaps that tradeoff isn't worth it. Wang et. al. at the University of Washington have recently done some excellent work on SPDY/3 that gives us a taste for this.[1] Eliot [1] https://www.usenix.org/system/files/conference/nsdi13/nsdi13-final177.pdf
Received on Monday, 25 November 2013 07:31:26 UTC