Re: [MMUSIC] RTSP for data recorder design

[Magnus Westerlund <[e-mail]>]

On 2013-09-04 18:19, Saylor, Kase J. wrote:
> Magnus,
> 
> Thank you so much for your reply. It would seem that RECORD is not
> function that really got much "traction"and I haven't been able to find
> any example of an RTSP server that provides the capability. Also, it
> would seem, that even though the RFC clearly states that RTP is not
> required, there doesn't seem to be a way to define a transport-protocol
> other that RTP. 
> 

I am not surprised that you are not fining implementations of RECORD.

Regarding alternative transport if you look at
https://datatracker.ietf.org/doc/draft-ietf-mmusic-rfc2326bis/ I hope
you will more clearly see how you can define alternative transports. In
RTSP 1.0 specification the extension points are not explicitly defined
nor have syntax that enables graceful handling of non-supporting agents.

To define a new transport is really the following:

1. Define the transport components that you need. For example IP/UDP/FOO
stack.
2. Give them an identifier to be used in the transport headers,
transport specs.
3. Determine which parameters your transport needs to exchange and how
that is accomplish in the transport header, possibly defining new
transport parameters.
4. Register the necessary parameters and identifiers to avoid collisions
with any other transport.

Cheers

Magnus


> --
> Regards,
> 
> Kase J. Saylor, PMP
> Manager - Tactical Networks &
> Communications
> Southwest Research Institute
> San Antonio, TX
> 210.522.3703
> [e-mail] <mailto:[e-mail]>
> http://tacticalnetworks.swri.org <http://tacticalnetworks.swri.org/>
> 
> On Sep 3, 2013, at 6:19 AM, Magnus Westerlund
> <[e-mail] <mailto:[e-mail]>>
> wrote:
> 
>> Hi,
>>
>> Looking at what you want to do I would say that RTSP could fit the bill.
>> However, the issue is the level of underspecification existing around
>> the RECORD usage of RTSP 1.0. We have left record out of RTSP 2.0
>> because there where no one to drive the use case. It can clearly be
>> defined as an extension.
>>
>> In RTSP my understanding the procedure to record a multicast stream the
>> recording server can receive would be the following. See example in 14.6
>> of RFC2326:
>>
>> 1. Client does ANNOUNCE with session description.
>> 2. SETUP for each stream to be recorded in same session where transport
>> header specifies that it is a multicast address to record from.
>> 3. Send RECORD to start recording.
>>
>> There is a lot of uncertainties from specification point of view how to
>> do this in RFC2326. RTSP 1.0 is 92 pages and RTSP 2.0 is 320 pages
>> without specifying recording, but adding some other features. This is an
>> example of how insufficient as a specification RFC 2326 is. If you think
>> RTSP is a good match, then perhaps building on RTSP 2.0 defining the
>> RECORDING functionality is the way for you.
>>
>>
>> Cheers
>>
>> Magnus
>>
>>
>> On 2013-08-08 21:44, Saylor, Kase J. wrote:
>>> Paul,
>>>
>>> Thanks for the link for siprec, unfortunately I am looking for a current
>>> IETF RFC that could be used. As siprec is informational at this time,
>>> it's not really an option. I think I'm pretty close with RTSP, but
>>> there's just a few things that I haven't sorted out. Perhaps I'll try to
>>> condense my confusing email into something a little bit more concise.
>>> Here we go:
>>>
>>> 1. I have data sources that provide data primarily as "streams" of
>>> multicast datagrams. We have specified the format of those datagrams,
>>> and we now need a way to record them so that they can be retrieved at a
>>> later time.
>>>
>>> 2. We have created a SOAP-based (role you own) interface that would
>>> provide the needed functionality, but if I could accomplish that
>>> functionality with RTSP, that would be better for the overall community.
>>>
>>> 3. I need the following functionality:
>>> a. Nominate certain (or all) multicast streams for recording (I assume
>>> ANNOUNCE would be the way to do that)
>>> b. Record said streams whenever needed
>>> c. Allow a client to find out what streams are available
>>> d. Allow the client to playback those streams
>>>
>>> So my first questions was about "adding" streams to a session. After
>>> much rambling and attempts to provide example RTSP communications
>>> between client and server, my last question was about how, using
>>> DESCRIBE, would I be able to show all the streams that were in a single
>>> session (based upon the assumption that multiple data streams could be
>>> group under one session).
>>>
>>> Does this at all help clear up my original post? Thanks again for your
>>> time and help.
>>>
>>> --
>>> Regards,
>>>
>>> Kase Saylor
>>> Southwest Research Institute
>>>
>>> On Aug 8, 2013, at 2:28 PM, Kase Saylor <[e-mail]
>>> <mailto:[e-mail]>
>>> <mailto:[e-mail]>> wrote:
>>>
>>>> Paul,
>>>>
>>>> Thanks for the link for siprec, unfortunately I am looking for a
>>>> current IETF RFC that could be used. As siprec is informational at
>>>> this time, it's not really an option. I think I'm pretty close with
>>>> RTSP, but there's just a few things that I haven't sorted out. Perhaps
>>>> I'll try to condense my confusing email into something a little bit
>>>> more concise. Here we go:
>>>>
>>>> 1. I have data sources that provide data primarily as "streams" of
>>>> multicast datagrams. We have specified the format of those datagrams,
>>>> and we now need a way to record them so that they can be retrieved at
>>>> a later time.
>>>>
>>>> 2. We have created a SOAP-based (role you own) interface that would
>>>> provide the needed functionality, but if I could accomplish that
>>>> functionality with RTSP, that would be better for the overall community.
>>>>
>>>> 3. I need the following functionality:
>>>> a. Nominate certain (or all) multicast streams for recording (I assume
>>>> ANNOUNCE would be the way to do that)
>>>> b. Record said streams whenever needed
>>>> c. Allow a client to find out what streams are available
>>>> d. Allow the client to playback those streams
>>>>
>>>> So my first questions was about "adding" streams to a session. After
>>>> much rambling and attempts to provide example RTSP communications
>>>> between client and server, my last question was about how, using
>>>> DESCRIBE, would I be able to show all the streams that were in a
>>>> single session (based upon the assumption that multiple data streams
>>>> could be group under one session).
>>>>
>>>> Does this at all help clear up my original post? Thanks again for your
>>>> time and help.
>>>>
>>>> --
>>>> Regards,
>>>>
>>>> Kase Saylor
>>>> Southwest Research Institute
>>>>
>>>>
>>>> On Aug 8, 2013, at 2:00 PM, Paul Kyzivat <[e-mail]
>>>> <mailto:[e-mail]>
>>>> <mailto:[e-mail]>>
>>>> wrote:
>>>>
>>>>> Kase,
>>>>>
>>>>> Its hard to discern your exact requirements. But you should at least
>>>>> look at the work in progress by the siprec workgroup. You can find
>>>>> the relevant drafts at:
>>>>>
>>>>> http://datatracker.ietf.org/wg/siprec/
>>>>>
>>>>> This may not be what you need, but is worth a look.
>>>>>
>>>>> Thanks,
>>>>> Paul
>>>>>
>>>>> On 8/8/13 8:49 PM, Saylor, Kase J. wrote:
>>>>>> To Whom it may concern,
>>>>>>
>>>>>> I am helping the US Army develop standards for a network recorder
>>>>>> on US
>>>>>> Army vehicles. We like to use available standards when at all
>>>>>> possible,
>>>>>> and my research seems to indicate that RTSP would be a great candidate
>>>>>> standard. I've been poring over the RTSP specification (along with the
>>>>>> SDP standard) and I have your douce code for LiveMedia. However, I
>>>>>> still
>>>>>> have a few questions concerning the RTSP standard and its use for our
>>>>>> particular need. It seems like there is plenty of examples on the
>>>>>> Internet for the PLAY method, but not much for RECORD. I really hope
>>>>>> that someone may be able to provide some insight that will get me
>>>>>> moving
>>>>>> in the right direction. On to the questions!
>>>>>>
>>>>>> Our recorder requires the ability to "nominate" data streams for
>>>>>> recording, so I believe the correct RTSP method is ANNOUNCE. My first
>>>>>> question is when I was to add a stream using ANNOUNCE, if I use
>>>>>> the same
>>>>>> Session number for multiple streams, does this basically create a
>>>>>> presentation? For example:
>>>>>>
>>>>>> C->S: ANNOUNCE rtsp://192.168.0.1/mission_recorder RTSP/1.0
>>>>>>          CSeq: 3
>>>>>>          Date: 7 Aug 2013 15:35:06 GMT
>>>>>>          Session: 47112344
>>>>>>          Content-Type: application/sdp
>>>>>>          Content-Length: 332
>>>>>>
>>>>>>          v=0
>>>>>>          o=- 2890844526 2890845468 IN IP4 192.168.0.123
>>>>>>          s=Threat Stream 1
>>>>>>          i=Threat stream from threat service on the Boomerang
>>>>>>          c=IN IP4 239.192.0.123/127
>>>>>>          t=0 0
>>>>>>          a=unique_id:threat1
>>>>>>  a=data_type:threat
>>>>>>          m=message 12345
>>>>>>
>>>>>> (some response back from server)
>>>>>>
>>>>>> C->S: ANNOUNCE rtsp://192.168.0.1/mission_recorder RTSP/1.0
>>>>>>          CSeq: 4
>>>>>>          Date: 7 Aug 2013 15:35:07 GMT
>>>>>>          Session: 47112344
>>>>>>          Content-Type: application/sdp
>>>>>>          Content-Length: 332
>>>>>>
>>>>>>          v=0
>>>>>>          o=- 2890844527 2890845471 IN IP4 192.168.0.123
>>>>>>          s=Position Stream 1
>>>>>>          i=Position stream from position service
>>>>>>          c=IN IP4 239.192.0.124/127
>>>>>>          t=0 0
>>>>>>          a=unique_id:pos1
>>>>>>  a=data_type:position
>>>>>>          m=message 12346
>>>>>> (some response back from server)
>>>>>>
>>>>>> Based upon these two announce messages, is it safe to assume that the
>>>>>> following command would cause the recorder to start recording
>>>>>> those two
>>>>>> data streams (assuming the current date/time is 8/4/2013 1530 GMT)?
>>>>>>
>>>>>>    C->S: RECORD
>>>>>> rtsp://192.168.0.1/mission_recorder/20130804T153000Z RTSP/1.0
>>>>>>          CSeq: 5
>>>>>>          Content-Type: text/parameters
>>>>>>          Session: 47112344
>>>>>>  Range: clock=20130804T153000Z-
>>>>>>
>>>>>> Furthermore, would the following command stop the recorder
>>>>>> (assuming the
>>>>>> current date/time is 8/4/2013 1532 GMT)? I didn't see anywhere in the
>>>>>> RFC for stopping a recording.
>>>>>>
>>>>>>    C->S: RECORD
>>>>>> rtsp://192.168.0.1/mission_recorder/20130804T153000Z RTSP/1.0
>>>>>>          CSeq: 6
>>>>>>          Content-Type: text/parameters
>>>>>>          Session: 47112344
>>>>>>  Range: clock=- 20130804T153200Z
>>>>>>
>>>>>>
>>>>>> I apologize but I have at least one or two more questions ;-) After
>>>>>> tearing down a session, how would one go back and playback recorded
>>>>>> data? From my RECORD example I've assumed that providing a URI with a
>>>>>> "path" based upon record start time, could be one way to mark
>>>>>> recording
>>>>>> events. Do you think this is an acceptable approach? Then the play
>>>>>> command would something like:
>>>>>>
>>>>>>    C->S: PLAY
>>>>>> rtsp://192.168.0.1/mission_recorder/20130804T153000Z RTSP/1.0
>>>>>>          CSeq: 6
>>>>>>          Content-Type: text/parameters
>>>>>>          Session: 1234567
>>>>>>
>>>>>> I suppose I could specify unique parameters and use the GET_PARAMETER
>>>>>> method to list the available recordings. Perhaps something like:
>>>>>>
>>>>>>    C->S: SETUP rtsp://192.168.0.1/mission_recorder RTSP/1.0
>>>>>>          CSeq: 1
>>>>>>  Transport: unicast;client_port=4588-4589
>>>>>>
>>>>>>    S->C: RTSP/1.0 200 OK
>>>>>>          CSeq: 1
>>>>>>  Date: 5 Aug 2013 11:15:07 GMT
>>>>>>          Session: 1234567
>>>>>>          Transport:
>>>>>> unicast;client_port=4588-4589;server_port=6256-6257
>>>>>>
>>>>>>    C->S: GET_PARAMETER rtsp://192.168.0.1/mission_recorder RTSP/1.0
>>>>>>          CSeq: 7
>>>>>>          Content-Type: text/parameters
>>>>>>          Session: 1234567
>>>>>>          Content-Length: 15
>>>>>>
>>>>>>  list_recorded_data
>>>>>>
>>>>>>
>>>>>>    S->C: RTSP/1.0 200 OK
>>>>>>          CSeq: 7
>>>>>>  Date: 5 Aug 2013 11:15:10 GMT
>>>>>>          Content-Length: 46
>>>>>>          Content-Type: text/parameters
>>>>>>
>>>>>>          list_recorded_data:20130804T153000Z,20130804T213000Z
>>>>>>
>>>>>>    C->S: TEARDOWN rtsp://192.168.0.1/mission_recorder RTSP/1.0
>>>>>>          CSeq: 8
>>>>>>  Session: 1234567
>>>>>>
>>>>>>    S->C: RTSP/1.0 200 OK
>>>>>>          CSeq: 8
>>>>>>
>>>>>>
>>>>>> Then we'd specify the proper way to interpret the response from the
>>>>>> server and open up a new RTSP session to playback the recorded
>>>>>> data. For
>>>>>> example:
>>>>>>
>>>>>>    C->S: SETUP
>>>>>> rtsp://192.168.0.1/mission_recorder/20130804T153000Z RTSP/1.0
>>>>>>          CSeq: 1
>>>>>>  Transport: multicast;client_port=4588-4589
>>>>>>
>>>>>>    S->C: RTSP/1.0 200 OK
>>>>>>          CSeq: 1
>>>>>>  Date: 6 Aug 2013 11:15:07 GMT
>>>>>>          Session: 7654321
>>>>>>          Transport:
>>>>>> multicast;client_port=4588-4589;server_port=6256-6257
>>>>>>
>>>>>>    C->S: PLAY
>>>>>> rtsp://192.168.0.1/mission_recorder/20130804T153000Z RTSP/1.0
>>>>>>          CSeq: 2
>>>>>>          Content-Type: text/parameters
>>>>>>          Session: 7654321
>>>>>>
>>>>>>    S->C: RTSP/1.0 200 OK
>>>>>>          CSeq: 2
>>>>>>  Date: 6 Aug 2013 11:15:10 GMT
>>>>>>
>>>>>> Okay, last question, would the DESCRIBE method for
>>>>>> rtsp://192.168.0.1/mission_recorder/20130804T153000Z be expected to
>>>>>> describe the "streams" that were recorded? In other words could the
>>>>>> following be expected?
>>>>>>
>>>>>> C->S: DESCRIBE rtsp://192.168.0.1/mission_recorder/20130804T153000Z
>>>>>> RTSP/1.0
>>>>>>          CSeq: 312
>>>>>>          Accept: application/sdp, application/rtsl, application/mheg
>>>>>>
>>>>>> S->C: RTSP/1.0 200 OK
>>>>>>          CSeq: 312
>>>>>>          Date: 7 Aug 2013 15:35:07 GMT
>>>>>>          Content-Type: application/sdp
>>>>>>          Content-Length: 332
>>>>>>
>>>>>>          v=0
>>>>>>          o=- 2890844527 2890845471 IN IP4 192.168.0.123
>>>>>>          s=Position Stream 1
>>>>>>          i=Threat stream from threat service on the Boomerang
>>>>>>          c=IN IP4 239.192.0.124/127
>>>>>>          t=0 0
>>>>>>          a=unique_id:pos1
>>>>>>  a=data_type:position
>>>>>>          m=message 12346
>>>>>>  s=Position Stream 1
>>>>>>          i=Position stream from position service
>>>>>>          c=IN IP4 239.192.0.124/127
>>>>>>          t=0 0
>>>>>>          a=unique_id:pos1
>>>>>>  a=data_type:position
>>>>>>          m=message 12346
>>>>>>
>>>>>> I really have no idea how multiple streams would be described!
>>>>>>
>>>>>> Thanks in advance for any help!
>>>>>> --
>>>>>> Regards,
>>>>>>
>>>>>> Kase Saylor
>>>>>> Southwest Research Institute
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> mmusic mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/mmusic
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> mmusic mailing list
>>>>> [e-mail]
>>>>> https://www.ietf.org/mailman/listinfo/mmusic
>>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> mmusic mailing list
>>> [e-mail] <mailto:[e-mail]>
>>> https://www.ietf.org/mailman/listinfo/mmusic
>>>
>>
>>
>> -- 
>>
>> Magnus Westerlund
>>
>> ----------------------------------------------------------------------
>> Multimedia Technologies, Ericsson Research EAB/TVM
>> ----------------------------------------------------------------------
>> Ericsson AB                | Phone  [phone]
>> Färögatan 6                | Mobile [phone]
>> SE-164 80 Stockholm, Sweden| mailto: [e-mail]
>> <mailto:[e-mail]>
>> ----------------------------------------------------------------------
>>
> 


-- 

Magnus Westerlund

----------------------------------------------------------------------
Multimedia Technologies, Ericsson Research EAB/TVM
----------------------------------------------------------------------
Ericsson AB                | Phone  [phone]
Färögatan 6                | Mobile [phone]
SE-164 80 Stockholm, Sweden| mailto: [e-mail]
----------------------------------------------------------------------