Slashdot Mirror
Internet Heading to Light Speed
dbaker writes "Wired Magazine has a very interesting article about the future of optical networks and the barriers we face before this technology is commonplace."
← Back to Stories (view on slashdot.org)
← Back to Stories (view on slashdot.org)
Can we have universal access and better content first please?
-Johan
PS> Oh yeah, contribute to wikipedia.
Do you really think that people are going to give up the ease of ethernet when your typical broadband maxes out at 1.5Mbit/s? I predict a painfully slow death of ethernet, which will probably go the way of the floppy drive.
Whats the point of blazing high speeds without the content???
..
Net's content value improvement rate is trending downwards
Optical-networking company Infinera is taking another approach... has developed a photonic integrated circuit, a hybrid of optical and electronic technologies... the technology combines discrete functions into a single chip, and can transmit data at speeds of up to 100 Gbps.
Although the 100Gb/s is the max, it would be interesting what the sustained rate would be.
This technology seems to have a better light/energy conversion than the 'bucky ball' solution, since it lists 40Gb/s as the transmission rate.
Perhaps the Inifera solution is limited in distance.
Superconnect's Lehenbauer agrees that "it's fascinating" to have material for an optical switch, but warns "it could be awhile until an all-optical network is possible." Lehenbauer said switches and routers must identify individual packets and route data intelligently, tasks that are not possible using a simple optical switch. "Unless you have an optical computer inside the switch to make these decisions, you'll still need electronic components."
Therein lies the bottleneck. Unless we develop optical computers (not for a while), we still need electronic switches and computers to analyze the content of the optical data in order to make intelligent decisions as to which direction the data should be channelled to.
Not to minimize the importance of this development, but until we do have optical computers, we are condemned to live life in the slow lane. But then again, someone may think of a clever way around this problem without using optical computers. One never knows.
I find it amusing that they are rolling these high speed services. If you have 50,000 people in a reasonably sized city all with 15mbps connectivity, do you really think they will all get that? I don't know how much ATM bandwidth is coming into any one CO, but I will bet it isn't 750,000mbps. Or better yet, wherever those DSL ATM connections terminate, I bet they don't have that much bandwidth available.
How is this any different than the broadband we have now?
Hundreds of cable modem users share a single 46mbit/sec downstream port on a CMTS. Do you think hundreds of customers on this port are all able to get 1.5, 3, and 6 mbits simultaneously? They don't.
And that is what providers, especially Verizon in this case, plan on.
And it has been proven, 95% of broadband customers don't use anywhere near their maximum bandwidth.
Bandwidth has had a high user to available bandwidth ratio for years. It's the only way the business is sustainable.
ISP's cannot afford to buy 1:1 bandwidth for every customer, unless we're going back to the 2400baud modem days.
1s. Minimal human decision time. Light travels 3e8m
1e-1s. Minimal human reaction time.
1e-2s. Minimal human recognition (sensory reaction) time.
1e-3s (1ms). Sensible task switching time.
1e-4s. in-task high level function time.
1e-5s. in-task medium level function time.
1e-6s (1us). Single microcontroller instruction; in-task low-level function time.
1e-7s Single high-speed microcontroller instruction.
1e-8s Single low-end CPU or DSP instruction time. Light travels 3m.
1e-9s (1ns) Single modern CPU time, light travels 0.3m
1e-10s A single modern CPU gate reaction time. Light travels 3cm, just above 1 inch.
Using standard $8 24bit ADC you can get down to the 3cm level with a $3 1MHZ microcontroller.
Using 1Gbit interface, your bits moving at light speed are 30cm away from each other.
A 300m LAN won't allow ping roundtrip shorter than 2 microseconds.
A 3000km (global network games) line WILL introduce perceptible delay.
A CPU of 3 GHZ just has to have its cache built in. Memory placed 3cm away causes 1 cycle long request-response roundtrip.
45 5F E1 04 22 CA 29 C4 93 3F 95 05 2B 79 2A B2
Your solution would require all-optical (what they call, "transparent") switch to re-configure itself on each of the packets that you are sending down. It's OK if one packet length is a complete ISO image, but is you are just sending 256 bytes to update your position in the game on other player's computer -- well, tough luck! ;-) Did you know that like a third of the packets on the Internet at any given time are under a couple hundred bytes long -- mostly TCP/IP ACKs.
Paul B.
I don't know exactly what you are thinking of, but I'm pretty sure the answer is no. "Splitting with a prism" is called Wave Division Multiplexing in fiber terminology. The number of frequencies that a single fiber can carry is partial dependant on the quality of the fiber, but mostly dependant on the devices transmitting and recieving the light. Regardless, there are a finite number of frequencies that can be utilized.
If you are suggesting to give each IP in the world it's own frequency - well, there are far to many address even in IP4 to do that with current technology. Even if we could it would be a huge waste of bandwidth, since most of the frequencies will be going unused most of the time.
Alternately, if you dynamically assign the N frequencies to N IP's that currently are sending data, then that will make efficient use of the available bandwidth, but when you get to the end of the fiber, the switch there doesn't know what IP corresponds to what frequency, so it would have to look at the data itself to do the switching. That is what this is doing and it is quite impressive.
Although I am by no means a router expert, it would seem logical that a majority of the latency in the network is caused by the actual reception and subsequent "analysis" of the packets. After all, the "response time", or "spped" of electricity is at best close to the speed of light.
This is a common misconception. Electrical transistors have a speed (read latency) that is mostly dependent on the voltage applied to them. (Hence over-volt on CPUs)
This technology would in effect change the rate that the transistor switched, thus changeing the latency.
Huh? Automatic telephone exchanges were pioneered by Almon B Strowger, who filed a patent in 1889. By 1922, the UK standardized its exchanges to use this system, eliminating human operators without using transistors.
UK telephone history