FastStart and payload size

Francois Audet audet at NORTELNETWORKS.COM
Fri Mar 10 14:55:17 EST 2000


Hum... That kind of worries me.

It means that the burden is on the originating side to "guess" what
the terminator is likely to support.

I also agree that practically speaking, implementations normally
support multiples of 10 ms as the payload size (with the notable
exeption of Netmeeting).

However, technically speaking, if you advertize 30 ms as the maximum
payload size you can support, it implies that in order to comply with
the H.323 specification, you would support any value between 0 ms
(whatever it means) and 30 ms. I don't think many implementations could
do this. But that is a separate issue.

In any case, if we all understand that the originator is responsible
for accurately describing a decent subset of the capabilities it can
support, we probably should describe this in the spec because I
am convinced that most people will simply put the maximum payload
size they can support...

> -----Original Message-----
> From: Paul Long [mailto:plong at SMITHMICRO.COM]
> Sent: Friday, March 10, 2000 11:41 AM
> To: ITU-SG16 at MAILBAG.INTEL.COM
> Subject: Re: FastStart and payload size
>
>
> Francois,
>
> You are correct that your first solution would cause
> backward-compatibility
> problems, because the called EP merely _selects_ which OLCs
> it wants to use.
> From the caller's proposed OLCs, it returns the ones that it
> accepts. It
> cannot provide its own OLCs or otherwise modify the proposed OLCs. The
> Recommendation is quite clear about this. I can quote from
> it, if necessary.
> :-)
>
> I'm afraid that the only solution is the second one you
> describe, but IMO it
> is not necessary to provide every possible value within a
> range, e.g., 100,
> 99, 98, 97... From my experience at interops, it seems like
> G.711 framing is
> limited to either 60, 30, 20, or 10ms frames, so that's what
> I would propose
> in addition to your own 100ms maximum. (Note that for some
> EPs, this is not
> their maximum framing but the _exact_ framing that you must use!)
>
> Our EP currently just proposes one type of media per session
> ID (due to
> another problem with FC). We propose G.723.1 and H.263. This
> is what our
> fastStart looks like in order to overcome the problem you describe:
>
> in G.723.1, maxframes 4, SS
> out G.723.1, maxframes 4, SS
> out G.723.1, maxframes 4, no SS
> out G.723.1, maxframes 1, no SS
> in H.263, SQCIF 1, QCIF 1, CIF 1, vTSTO
> out H.263, SQCIF 1, QCIF 1, CIF 1, vTSTO
> out H.263, QCIF 1, vTSTO
> out H.263, QCIF 2, no vTSTO
>
> Note that it's only necessary to provide maximum caps for
> incoming media,
> because the called EP can always ignore features it does not
> support, e.g.,
> SS and CIF, and undershoot your maximum framing. Once/if we
> support multiple
> media per session ID, this is probably what it would look like:
>
> in G.723.1, maxframes 4, SS
> out G.723.1, maxframes 4, SS
> out G.723.1, maxframes 4, no SS
> out G.723.1, maxframes 1, no SS
> in G.711, ALaw, maxFrames 200
> out G.711, ALaw, maxFrames 60
> out G.711, ALaw, maxFrames 30
> out G.711, ALaw, maxFrames 20
> out G.711, ALaw, maxFrames 10
> in G.711, muLaw, maxFrames 200
> out G.711, muLaw, maxFrames 60
> out G.711, muLaw, maxFrames 30
> out G.711, muLaw, maxFrames 20
> out G.711, muLaw, maxFrames 10
> in H.263, SQCIF 1, QCIF 1, CIF 1, vTSTO
> out H.263, SQCIF 1, QCIF 1, CIF 1, vTSTO
> out H.263, QCIF 1, vTSTO
> out H.263, QCIF 2, no vTSTO
> in H.261, QCIF 1, CIF 1, vTSTO
> out H.261, QCIF 1, CIF 1, vTSTO
> out H.261, QCIF 1, vTSTO
> out H.261, QCIF 2, no vTSTO
>
> (whew!)
>
> Paul Long
> Smith Micro Software, Inc.
>
> -----Original Message-----
> From: Francois Audet [mailto:audet at NORTELNETWORKS.COM]
> Sent: Friday, March 10, 2000 12:35 PM
> To: ITU-SG16 at MAILBAG.INTEL.COM
> Subject: FastStart and payload size
>
>
> Another glitch with fast connect...
>
> The OpenLogicalChannel indicates in dataType.audioData the MAXIMUM
> payload size for the forward channel to be opened. In non-fast start,
> the endpoint opening the channel would obviously use the
> information it
> gathered in the terminal capability exchange in order not to
> try to open
> a channel that would exceed's the receiver's capabilities.
>
> In fastStart, there is no terminal capability negotiation.
> The endpoint
> sending the SETUP message includes the MAXIMUM payload size it can
> support in both direction since it can not guess what are the
> capabilites of the receiver. The receiver has to "select"
> from the list
> provided in the SETUP.
>
> Suppose that A calls B using fast start. A has a big buffer and can
> support G.711 at up to 100 ms (send and receive). B can only
> support 20
> ms. A calls B using fastStart and advertizes 100 ms for both
> direction.
> B has no choice but to reject the call, or to initiate H.245
> tunnelling
> or a separate H.245 channel. This means to me that it can
> actually be a
> disadvantage to advertize a bigger maximum payload size than
> the default
> in fasStart. It also mean that fast start only has a reasonable chance
> of working well if by chance the called party can handle an equal or
> larger payload size than the originator.
>
> Any views on this?
>
> One possibility is that the receiving endpoint would be allowed to
> signal a SMALLER payload size in the OLC it sends back to the
> originator. This would however mean a change to the specification, and
> would potentially mean backward interoperability problems
> since current
> implementations may not accept OLC contents that do not match what was
> originally sent in the OLC in the SETUP message (they might reject the
> call, or might ignore the changed values).
>
> The other possibility is for the sender of the SETUP message to
> advertize multiple payload size as alternative OLCs. While it would be
> possible in some cases to completely describe one's
> capabilities without
> creating huge fastStart messages (eg., G.723-30 ms, G.723-60
> ms), it may
> not be practical for others (e.g., G.711-100 ms, G.711-99 ms, G.711-98
> ms, ... down to 1 ms). This would mean that in some cases,
> you might be
> restricted to only a few (maybe advertize the maximum and the
> default).
> Maybe a trick would be to advertize always at least the
> default (20 ms)
> if supported, and the maximum if different.
>
> In any case, I think we should clarify the procedures.
>
> -----
> François Audet  Tel:+1 408 565 5675 mailto:audet at NortelNetworks.com
> <mailto:audet at NortelNetworks.com>
> Nortel Networks Fax:+1 408 565 2375 http://www.NortelNetworks.com
> <http://www.nortelnetworks.com/>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.packetizer.com/pipermail/sg16-avd/attachments/20000310/0cce6890/attachment-0004.html>


More information about the sg16-avd mailing list