Re: [pntaw] Real-time media over TCP

On Oct 14, 2013, at 2:42 PM, Paul Kyzivat <pkyzivat@alum.mit.edu> wrote:

> On 10/14/13 3:06 PM, Markus.Isomaki@nokia.com wrote:
>> Hi,
>> 
>> In practice I doubt you find many situations where UDP is completely blocked but incoming TCP connections from anywhere are allowed. I'm not sure the use cases and requirements document really claims that to be the case, but I agree it is somewhat ambiguous. That means that while direct TCP connection would be better than TCP based relay, its success rate would be very small. I suppose we agree that UDP based relay would still be better than direct TCP for real-time media.
> 
> Doesn't that depend on where the relay is?
> 
> UDP relayed across the world might not be preferable to TCP direct to a point down the street.
> 
> But in practice I suppose the relay will be conveniently located relative to one of the two endpoints.

I think that is a separate problem.  Smart ICE candidate prioritization would choose the "best candidate" (for some value of "best candidate").  The ICE connectivity check round-trip time is an oft-cited example of a reasonable metric (although it doesn't measure bandwidth and packet loss is considered by some to mislead [but if a path is lossy it may indeed be a very undesirable path]).  Geolocation of the TURN IP address versus my own IP address could also influence ICE candidate prioritization or candidate selection.

-d

> 
> 	Thanks,
> 	Paul
> 
>> Markus
>> 
>>> -----Original Message-----
>>> From: pntaw-bounces@ietf.org [mailto:pntaw-bounces@ietf.org] On Behalf
>>> Of ext Parthasarathi R
>>> Sent: 14 October, 2013 21:55
>>> To: pntaw@ietf.org
>>> Subject: Re: [pntaw] Real-time media over TCP
>>> 
>>> Hi all,
>>> 
>>> My point is that the direct media connection has to be given priority over
>>> TURN based mechanism. In case of ICE-TCP, there is a possibility for the
>>> direction connection between two browsers involved in the WebRTC session
>>> and should be given priority over TURN based mechanism. So, 6) TCP based
>>> candidates (ICE-TCP) - MUST
>>> 
>>> There is an assumption in the discussion that the incoming TCP traffic is not
>>> allowed through firewall. In case it is the problem to be solved, RTCWeb
>>> usecase and requirement has to be updated. I noticed in
>>> draft-ietf-rtcweb-use-cases-and-requirements-12 (published today) that
>>> there is no such requirement. I'll write the mail in RTCWeb WG to get the
>>> clarity on the requirement in case the firewall forbidding incoming TCP traffic
>>> is a matter of missing text in the requirement.
>>> 
>>> Thanks
>>> Partha
>>> 
>> _______________________________________________
>> pntaw mailing list
>> pntaw@ietf.org
>> https://www.ietf.org/mailman/listinfo/pntaw
>> 
> 
> _______________________________________________
> pntaw mailing list
> pntaw@ietf.org
> https://www.ietf.org/mailman/listinfo/pntaw