[opendtv] Re: Netflix's Move Onto the Web Stirs Rivalries
- From: Craig Birkmaier <craig@xxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Mon, 29 Nov 2010 07:32:52 -0500
At 5:53 PM -0600 11/28/10, Manfredi, Albert E wrote:
Can you describe exactly what "improvements in streaming technology"
you think have been so instrumental in this, since 2003?
There have been a number of advances in this area.
First, most of the current streaming media servers have been
developed since 2005, including Flash, QuickTime streaming server,
Real, Microsoft et al. They employ a variety of techniques to improve
performance and reliability. This site has some good reference
material and studies of performance conducted between 2006 and 2010:
http://www.iupui.edu/~nmstream/
Channel Adaptive streaming techniques have contributed to performance
improvements in several areas:
1. Client control of server playout rates
2. Rate distortion optimized packet scheduling
3. source encoder channel-adaptive packet dependency control
This 2002 paper from researchers at Stanford University discusses the
techniques listed above:
http://www.stanford.edu/~bgirod/pdfs/GirodICIP02invited01.pdf
Also, continuous updates to Internet routers with support for RTSP
and other new standards have contributed. And significant growth in
distributed server platforms like Akamai have moved much of the media
that is being served closer to the edges of the network.
> The principal cause of the problem is the architecture of first-decade
technology;
Can you explain exactly what you are calling "architecture of
first-deacde technology"?
Not my words. I believe Peter was discussing the generic use of TCPIP
one-to-one sessions to deliver content to consumers, and the inherent
inefficiencies of this approach to broadcasting (and by implication
IP Multicast). This was the same paragraph in which Peter discussed
the notion of periodically broadcasting popular content to cache,
which I have labeled Data Broadcasting.
MPEG-2 is now wastes huge amounts of bandwidth for both broadcasters
and the MVPDs.
Same old non-credible mantra. "Huge amounts" my foot. In spite of
the rhetoric, when it comes to quoting actual bit rate requirements
used in transmitting TV quality SD and HD, it isn't H.264 that
"saved the day" at all. The bit rates are very close between the
two, EVEN IF the quality achieved at a given bit rate might show
gains with H.264.
Sorry, but h.264 streaming rates are typically 30 to 50 percent less
than the equivalent quality with MPEG-2. It is true things begin to
even out at higher bit rates, such as those used in broadcast
channels. But for streaming you will not see much MPEG-2.
On the other hand, when more people used dialup than broadband for
Internet access, dialup being no better than 52 Kb/s downstream
(even though it was called 56K) and broadband in the neighborhood of
768 Kb/s downstream, it was very obvious that decent quality movies
were not going to become available streamed in real time.
I have had broadband since the late '90s. THe quality of streaming
sucked until the middle of the past decade, and has improved
dramatically in the past 2-3 years. ONly a small portion of this is
due to improvements in delivered bit rates from my AT&T DSL service.
The one overriding difference is the speed of the last mile
connection. With the current 10 to 50 Mb/s downstream broadband
connections you can get from either cable or FiOS, streaming of HD
quality movies in real timer is VERY feasible, even with MPEG-2
compression, Craig. Without having to rely on vague and
unsubstantiated references to "architecture of first-decade
technology," Netflix has no problem selling more streamed movies
than DVD rentals, right about now.
Few people in the U.S. have broadband in the 10 - 50 Mbps range, and
when they do it is typically associated with a service like Fios,
which has dedicated "channels" for delivering HD content. Most HD
streaming services are optimized for bit rates of 4-5 Mbps.
I would also note that consumers have become accustomed to crap since
the transition to DTV. We tolerate compression artifacts and other
distortions that can be quite severe, because on average the pictures
are better than analog TV. This is true for broadcast TV, the MVPDs,
and streaming. At least with streaming there are some services that
focus on delivering a consistently good viewing experience, albeit
with the need for significant buffering (or pre-downloading).
Regards
Craig
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:
- Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.
Other related posts:
