On 02/24/11 15:43, Lars Eggert wrote:
Hi,
the charter makes sense to me overall.
I have a minor question about:
On 2011-2-24, at 15:44, Harald Alvestrand wrote:
5) Diffserv based QoS What's envisioned here? AFAIK, DSCPs are being ignored and/or rewritten pretty much universally, so relying on them for some required functionality is going to be problematic. Using DiffServ as an optional optimization to gain some additional benefits across those few paths where it is actually enabled is of course not a problem.
(FWIW, none of the rtcweb IDs seem to mention DiffServ at the moment...) YMMV; my interpretation is that we need to discuss whether or not browsers need to offer functionality for "send this stream with DSCP X".
The relevant questions are: - Is it important enough to mandate? - Is the determination of useful values of "X" possible to do based on available information? A perfectly valid conclusion for the WG to come to is "no, it's not worth it". I have been told that people who deploy VoIP in constrained environments (corporate networks, 3/4G deployments) find setting DSCP useful, but they don't depend on generic PCs attached to the network doing it "right", instead they do it through middle boxes that tweak DSCP values based on configuration, provisioning protocols or deep packet inspection. Others who know of specific instances can certainly speak more. Harald
Lars