[RTW] [dispatch] Charter proposal: The activity hitherto known as "RTC-WEB at IETF"

[Harald Alvestrand]

On 01/18/11 11:15, Stefan Håkansson LK wrote:
> I agree to that there are situations when a fallback to http (or perhaps WebSockets for a more efficient transport!) is required.
>
> But I don't think the http streaming solutions are the answer. Those solutions to my understanding build on chopping up a stream into segments, and each segment is represented by a file. Each segment is typically several seconds long, and that is not sufficient for conversational services where the mouth-to-ear delay should be far below 1 second (yes, we should discuss if there should be a requirement, and what it should be in that case).
>
> Sure you could use much shorter segments, but with all the encapsulation/wrapping data the overhead would get really big.
There are multiple HTTP streaming options. Let's not confuse ourselves 
with saying "HTTP streaming" without referring to a specific proposal - 
they're just too different.

(and mea culpa - I need to do that too...)

>> -----Original Message-----
>> From: Harald Alvestrand [mailto:harald at alvestrand.no]
>> Sent: den 17 januari 2011 19:33
>> To: Marshall Eubanks
>> Cc: Bernard Aboba; Stefan Håkansson LK; tom.taylo at huawei.com;
>> markus.isomaki at nokia.com; rtc-web at alvestrand.no; dispatch at ietf.org
>> Subject: Re: [dispatch] [RTW] Charter proposal: The activity
>> hitherto known as "RTC-WEB at IETF"
>>
>> Another way (that I know is in production) is to use RTP with
>> a thin shim layer (length fields) to provide packet
>> separation, straight over TCP. If we are certain we have to
>> support RTP-over-UDP, that might be the solution that has the
>> lowest extra implementation cost over the UDP solution.
>>
>> But I suspect this will have to  be decided based on a set of
>> requirements, not just a beauty contest.
>>
>> What is the added value of the MPEG-2 wrappers?
>>
>> On 01/17/11 18:23, Marshall Eubanks wrote:
>>> On Jan 17, 2011, at 10:59 AM, Bernard Aboba wrote:
>>>
>>>> +1.
>>>>
>>>> One place where we could "spend our energy wiser" might be
>> on enabling interoperability
>>>> of HTTP transported realtime media.   Although
>> peer-to-peer traffic is more desirable when
>>>> possible, "HTTP fallback" is in practice required a significant
>>>> fraction of the time, due to the prevalence of highly
>> restrictive firewalls.
>>> I would agree, and that raises the issue of the "wrapper"
>> for HTTP streaming. Note that Apple uses MPEG-2 TS for the
>> wrapper for its live http video streaming.
>>> (  Each media file MUST
>>>      be formatted as an MPEG-2 Transport Stream, an MPEG-2
>> Program Stream,
>>>      or an MPEG-2 audio elementary stream  -
>>> http://tools.ietf.org/html/draft-pantos-http-live-streaming-01 )
>>>
>>> While this is certainly standards based, I do not think it
>> matches or interoperates with anyone else's HTTP streaming.
>> And, of course, this is an I-D still. Flash also does http
>> streaming, but I believe it uses its own, proprietary, wrapper.
>>> So, is specifying a media transport protocol for http
>> streaming in scope ?
>>> Regards
>>> Marshall
>>>
>>>
>>>>> From: stefan.lk.hakansson at ericsson.com
>>>>> To: tom.taylo at huawei.com; harald at alvestrand.no;
>>>>> Markus.Isomaki at nokia.com
>>>>> Date: Sat, 15 Jan 2011 15:17:51 +0100
>>>>> CC: rtc-web at alvestrand.no; dispatch at ietf.org
>>>>> Subject: Re: [RTW] [dispatch] Charter proposal: The
>> activity hitherto known as "RTC-WEB at IETF"
>>>>>> I agree that at least for the time being it is more fruitful to
>>>>>> focus the energy elsewhere. There is plenty of useful
>> work that can
>>>>>> be done about media transport (the datagram service and the
>>>>>> potential bytestream
>>>>>> ) and the associated APIs, and I suggest we focus on
>> that. We can
>>>>>> try our luck with the codec thing later on.
>>>>> I agree. Codec discussions seem to go on forever, and we
>> could spend
>>>>> our energy wiser.
>>>>>
>>>>> Stefan
>>>>>
>>>>> PS Sorry for answering late, but I did not follow dispatch. I
>>>>> thought all related messages would go on rtc-web as well.
>> So those
>>>>> of you who do not follow dispatch: perhaps you should
>> look into the dispatch archive.
>>>>> _______________________________________________
>>>>> RTC-Web mailing list
>>>>> RTC-Web at alvestrand.no
>>>>> http://www.alvestrand.no/mailman/listinfo/rtc-web
>>>> _______________________________________________
>>>> dispatch mailing list
>>>> dispatch at ietf.org
>>>> https://www.ietf.org/mailman/listinfo/dispatch
>>