re: videostreaming over cablemodem plant
I picked up the snip below on the Society of Cable Telecom Engineers (SCTE) mailing list. It's a post that was made in response to a cable engineer's question concerning download speeds and congestion on cable modem HFC plant, and the collective effects of these phenomena on videostreaming applications. I thought it interesting enough to post here.
Note the engineer's incorrect use of the term "MB," which stands for MegaBytes, when in fact the intention was megabits per second, or "Mb/s."
I'm not in any way suggesting that the assessments presented below by this engineer are accurate or even technologically sound. I find that many Cable TV engineers - and this goes for many ISP engineers, as well - with otherwise impeccable qualifications in their own idioms find themselves completely out of their elements when addressing digital disciplines, especially those that center on common carrier technologies and nomenclature.
I find no fault with that, except when they profess to know better than "you" what they are talking about, where "you" is anyone with a real grasp of carrier disciplines. And this works in the reverse, as well.
Rather, the value of the post might be to demonstrate how a particular mindset with a cabletv perspective approaches issues such as videootreaming and other bandwidth intensive applications over CM facilities. Enjoy.
FAC
---begin:
[greetings,]
My experiments with MPEG2 video over cablemodems have been mostly
unsuccessful. I believe that bandwidth is the primary limit. I'm sending
this message via a cablemodem at home. The University of Wisconsin-Madison
has a project with Charter Business Networks to provide cablemodem service
for UW students, faculty and staff. Traffic from these cablemodems is
delivered over a direct fiber digital connection right back to our campus.
However, Charter has limited the bandwidth of the cablemodems to 1.5 MBPS.
Interestingly, a demonstration of 5 MBPS MPEG2 video worked just fine over
an 8 MBPS aDSL line at a hotel.
In my opinion, it won't be practical to send video over cablemodem IP until
bandwidths exceed 8-10 MB. There is also the problem of the shared
bandwidth at the cable company node - all the cablemodems in the node share
the downstream bandwidth. So the idea of hundreds of cablemodem
subscribers viewing video over cablemodem IP is not practical.
Here at UW, we would like to deliver cable TV-like video over IP to campus
users, but even in our campus environment, there are a number of
departments who have shared Ethernet hubs, and IP video doesn't work well
in those environments. We are kicking around plans to upgrade our current
campus OC-12 ATM backbone to switched Gigabit Ethernet. That should
support video over IP.
Our College of Engineering and Computer Science Department have piloted an
application they call eTeach. You can read about it at
<http://eteach.engr.wisc.edu/>. It supports multiple synchronized windows
with Windows Media Player, Powerpoint slides, and World Wide Web
pages. They have converted their CS310 class to eTeach lectures so
far. The problem is in getting all the clients running the correct version
of Windows with enough memory and CPU speed, and the correct version of
Windows Media Player. Also, viewing is only supported in labs with
switched 100 MBPS Ethernet.
WHA, our local PBS station is working with PBS on a distributed video
editing system, allowing an editor in one city to integrate video clips
from other stations in other cities. This will use the Internet2 Abilene
network.
Last year we shipped MPEG2 video of a live hockey game from Madison to
Anchorage where it was broadcast by the public station up there. This year
we're trying to exchange live women's basketball MPEG2 video over Abilene
between Wisconsin, University of Indiana and Penn State.
I'm working with a number of people on a new public television and radio
network in the State of Wisconsin. The network will use MPEG2 video over
an ATM network. Why not IP? First, this network will not be shared, so
there are no other users to share the cost of the network. Second, the
network will distribute one television and two radio networks, and IP has
no quality of service feature. Third, the budget will only support DS-3
bandwidth, which doesn't give IP video very much headroom. The only way we
can do this network with QOS is with ATM.
[signed]
---end |
