Slashdot Mirror
Next-Gen Broadband Primer
Aaron writes "Broadband Reports has a good read on the real deal behind next generation broadband deployments. In four years: half all Verizon DSL users should have fiber, half of all SBC subscribers should have 10-20Mbps DSL, and one tenth of all BellSouth customers should have 50Mbps DSL. At the same time cable companies should begin deploying DOCSIS 3.0 technology in 2006, eventually bringing 100Mbps speeds to end users."
nothing like bandwidth better than most of the servers you connect too..... You would practically cause a DOS attack with prefetch alone.....
Rural America is fun fun fun.
Now, what about latency and QoS?
And there was way too much mention of IPTV and you-know-who, with their "the future may run through us alone" attitude, in that article for it to be palatable.
One little mention of broadband over the power lines (BPL)?
Interesting since Google just made a huge investment in it.
Execute? [Y/N] _
end users always seem to not be able to understnad this. There's only "cost" with a DS3/OC-3 because you're leasing it from someone else. These big ISP's own their own pipes, then have peering agreements with other providers. They don't have to "buy" a DS-3, they only have to slap some hardware on either end of their fiber to make it a faster pipe. It doesn't cost them "more" to up end-user speed unless they're breaking peering agreements, which isn't likely.
It's possible for a product to improve while the "real cost" remains the same. Why should broadband connections be different than anything else?
They shouldn't but considering that broadband connections have gotten *slower* while costs have risen (i.e. AT&T@Home (up to 10mbit) -> ATTBI (1.5mbit)), people really shouldn't believe this round of hype.
What I'd like to see is better utilization of bandwidth within the cable/DSL network infrastructure. It costs cable/DSL providers MUCH less to provide high speed connections between customers in the same local topology.
If you had 100Mb/s to everybody within your local area it would make things like high speed videoconferencing or sharing of high bandwidth content between friends and family VERY fast.
Problem is that current caps on cable/DSL lines dont' descriminate between transfers between two people on the same cable/fiber segment and going out beyond the border router and down that T3/OC-3 or whatever out to the commercial internet up provider. As a result you are capped at communicating with the person accross the street when you really could communicate with them at blazing speeds.
Read this 1999 article about SBC's 'Project Pronto'. " According to SBC, when the expanded deployment program is completed [in three years] customers will be able to receive minimum downstream connection speedsof 1.5 megabits per second, with more than 60 percent eligible to receive guaranteed speeds of 6 megabits a second." Right.
SBC's new "Project Lightspeed" isn't about the Internet at all. It's just cable TV, implemented using Windows Media 9 over DSL using Scientific-Atlanta set-top boxes. The system doesn't use the Internet at all. It has its own infrastructure, which is a Microsoft-implemented multicast implementation.
It's not about Internet access at all. All you can get is what they want to send you. Lightspeed will block access to Internet video.
Imagine trying to play Quake on a remote system.
That's pretty much how it works now. When you play a network game, you might as well be running on an OpenGL-tweaked X terminal.
Client-side prediction is helpful for a good experience in the general case, but it's far from necessary on most broadband connections today, and it won't be necessary at all in the future.
No matter how much bandwidth you've got, the game just won't run right with 1/3 second lag. Then, also assume that your "real computer" has to talk to another server. It'll be like playing in mud.
No, it won't be substantially different from the way it works now on a non-client-side-predicted client like the original Q1. You don't see the effect of moving your mouse or hitting a key until it goes to the server and back. All that will change is where the resulting graphical view is rendered. There is no inherent reason why the gameplay experience should feel any different, as long as you have enough downstream bandwidth to accomodate a video stream.
I'm not claiming any obvious gameplay advantages in a scheme like this, just pointing out that there's no reason why it wouldn't work. (Certainly it would eliminate cheating, since no geometry would ever exist on the client....) There are some interesting business aspects to the idea, such as the fact that instead of buying a $400 GeForce 68000 card or whatever, you're basically renting time on a real-time render farm somewhere.
Dahlmann tightly grips the knife, which he may have no idea how to use, and steps out into the plain.