• Quick note - the problem with Youtube videos not embedding on the forum appears to have been fixed, thanks to ZiprHead. If you do still see problems let me know.

Software breaks data transfer record?

knightmb

New Blood
K
Joined
Dec 27, 2001
Messages
9
http://www.news24.com/News24/Technology/News/0,,2-13-1443_1350339,00.html
Pasadena - US scientists have broken the record for the top speed for transferring data over the internet.

The magazine Nature reported in its online edition that a new piece of software allows data to be transferred up to 3 500 times faster than normal over broadband connections.

That type of speed can ultimately make it possible to download a DVD film from the internet in fewer than five seconds.

"We hope that we will be able to publish a first version of the software this summer," said Steven Low, one of the researchers.

Low and other scientists from the California Institute of Technology in Pasadena hope to offer the software for free over the internet.

The "bottleneck" in the transfer of data over the World Wide Web comes from the Internet protocol TCP, the report claims. When errors emerge during the transfer of data packets, the protocol presumes that the network is overloaded and halves the data transfer rate immediately.

"The remedy is much too drastic," explains Low, even if it was a terrific idea back in the late 80s when it was implemented.

His software measures the delay in sending the data and calculates the workload based on that figure. If the delay rate climbs, the workload is reduced only slightly. - Sapa-DPA
Click to expand...

I can see maybe a double in bandwidth, but I believe the claim of 3,500 times faster is pure bunk in my opinion. I'm having a hard time explaining to the rest of the tech population. Anyone care to bring me up to date why this wouldn't be possible? Even some common sense examples would be great.
 
Hmm.
A burst of noise can cause quite a few halvings of rate all at once. A burst of noise is pretty much guaranteed to happen. 3500, though, is more than 2^11.
No matter how you parse it, Claude Shannon's theorems apply to the Internet the same as to any noisy pipe. Look up that name and you may find the math needed to show that such astounding rates are just not possible with the current infrastructure.
Now, I could see using, say, 100 telephone lines to accomplish a 100-fold data transfer increase. Tricky, but possible....
Internet backbones already move data at rates comparable to a DVD ever 5 seconds... but that data flow is demultiplexed to end users...
 
Neat. This link seems to explain it a little better:

http://www.nrilinks.com/news/n/12/4771.aspx

It seems to me they are comparing apples and oranges. Their data link was capable of sustaining 923mbits/sec, and they are comparing that to a "typical broadband connection", which in my experience is more like <2mbits/sec (my cable modem usually gets about 150KBytes/sec on a good day, which is roughly 1.5mbits/sec).

I note that if I divide 923mbits/sec by 3500, I get about 260kbits/sec, which is somewhere between a dial-up modem's 56kbits and cable modem's 1500kbits. Maybe some DSL links are about this speed? Don't know.

It just seems strange to say that a piece of software is responsible for this speed difference, which is totally incorrect. You need the raw bandwidth in hardware. It sounds like something was lost in the translation from the techie folks to the copy writers. I'm sure they are trying to say that their software makes more efficient use of the available bandwidth than TCP/IP does, which I can belive. But a piece of software absolutely will not ever make currently available cable modems run at 923mbits/sec.

I think I've narrowed it down to this highly ambiguous and very poorly worded sentence: "The magazine Nature reported in its online edition that a new piece of software allows data to be transferred up to 3 500 times faster than normal over broadband connections." If you read that quickly (or even slowly a couple times), you might come away thinking that a piece of software can boost a cable modem 3500 times faster. But that's not at all what actually happened.

Note also the article mentions that this speed is faster than is usable by current PCs.

Hope this helps.
 
You must log in or register to reply here.

Back
Top Bottom