[foms] Proposal: adaptive streaming using open codecs

[Silvia Pfeiffer]

On Tue, Oct 19, 2010 at 11:53 AM, Michael Dale <dale at ucsc.edu> wrote:
>  On 10/18/2010 04:58 AM, Jeroen Wijering wrote:
>> Hello all,
>>
>> Here is a (rough and incomplete) proposal for doing adaptive streaming using open video formats. WebM is used as an example, but all points should apply to Ogg as well. Key components are:
>>
>> * Videos are served as separate, small chunks.
>> * Accompanying manifest files provide metadata.
>> * The user-agent parses manifests and switches between stream levels.
>> * An API provides QOS metrics and enables custom switching logic.
>>
> Just to quickly add in my 2c.
>
> I don't see much advantage in splitting everything into chunks. At least
> for video on demand, the easiest to use system would just extend the
> existing <video> tag embed syntax with some additional streams. You
> would lose a slight bit of efficiency as you download the index header
> for each stream to support stream switching.. but that should be
> compared to the efficiency of downloading the manifest file over and
> over.  And with delta encoded indexes, the key frame indexes are not
> necessarily very large.
>
> Of course for live streams the chunk approach makes more sense so
> probably have to support both.

You don't have to load and re-load the manifest for non-live streams,
so that part would go away.

As for the actual chunking - in theory a server component should be
able to do the chunking on the fly where it's not live - similar to
what oggz-chop does. So it's possible to do that without requiring the
Web developer to prepare the chunking where only one Web server is
involved.

The advantage of the chunks it that the CDN providers already
understand how that works for mpeg and can support it the same way for
other formats. I think that's a big advantage.

Silvia.