Posted: 5/26/2014 2:04:11 PM EDT
|
|
Quoted:
AT&T has it together they just bought the satelite business,... The future is in the sky not under water.  not according to my bandwidth-delay product calculator. http://en.wikipedia.org/wiki/Bandwidth-delay_product ar-jedi |
|
Quoted:
Anyone remember when they tried to cut the trans-pacific cable on Gilligan's Island?
http://1.bp.blogspot.com/-5rEgh1dwKMo/Uv5ZbvUyk-I/AAAAAAAAEFM/oUAOLyk9ghM/s1600/TORCH.png Man, I'm glad I was a kid when I used to watch that show. Because I'm not sure how that cable was re-routed all the way down to the Bahamas. |
|
LOL I must have missed that episode MAP... http://www.submarinecablemap.com/ |
Back when I lived in Key West, and before the advent of cell phones, about once a year someone would manage to cut the only phone cable that ran down the Keys. Typically, that would mean you could only place a call from Key West to somewhere like Marathon. And if the drawbridge got stuck in the "up" position, you were completely hosed. 
But then, being stuck in Key West, and isolated from the rest of the world was actually kind of nice. 
|
|
Quoted:
Man, I'm glad I was a kid when I used to watch that show. Because I'm not sure how that cable was re-routed all the way down to the Bahamas. Quoted:
Quoted:
Anyone remember when they tried to cut the trans-pacific cable on Gilligan's Island?
http://1.bp.blogspot.com/-5rEgh1dwKMo/Uv5ZbvUyk-I/AAAAAAAAEFM/oUAOLyk9ghM/s1600/TORCH.png Man, I'm glad I was a kid when I used to watch that show. Because I'm not sure how that cable was re-routed all the way down to the Bahamas. They left port from Honolulu |
|
Quoted:
When they're wrapped in layer after layer of plastic, then steel, then rubber, I imagine so. Pretty damn interesting that they used vacuum tube amps, I never would have guessed that! Quoted:
Quoted:
How deep was it? Vacuum tubes can take that kind of pressure? When they're wrapped in layer after layer of plastic, then steel, then rubber, I imagine so. Pretty damn interesting that they used vacuum tube amps, I never would have guessed that! you'll find this interesting: http://hackaday.com/2013/12/24/retrotechtacular-submarine-cable-splicing-is-serious-business/ ps regarding powered midspan amps... modern fiber optic submarine cable spans benefit from three things: increased receiver sensitivity, coding gain due to FEC (which can be quite high -- 6 to 10dB of effective gain!) , and Raman distributed amplification. using these techniques, unrepeated spans of up to 400KM are possible. for longer spans, inline EDFA and Raman amplifiers are used, powered by several KV of DC supplied by the end stations. both amplifier types operate irrespective of the underlying protocol, data rate, and support multi-wavelength operation (DWDM) as well. the net result of this is that today a single fiber pair can carry up to 128 channels (lambdas) of 100Gbps traffic, or about 12.8 terabits per second. already, 200Gbps and 400Gbps end terminal equipment is in trial, and this will double or quadruple the span capacity. http://en.wikipedia.org/wiki/Forward_error_correction http://en.wikipedia.org/wiki/Raman_amplification http://en.wikipedia.org/wiki/Wavelength-division_multiplexing ar-jedi |
|
Quoted:
you'll find this interesting: http://hackaday.com/2013/12/24/retrotechtacular-submarine-cable-splicing-is-serious-business/ ps regarding powered midspan amps... modern fiber optic submarine cable spans benefit from three things: increased receiver sensitivity, coding gain due to FEC (which can be quite high -- 6 to 10dB of effective gain!) , and Raman distributed amplification. using these techniques, unrepeated spans of up to 400KM are possible. for longer spans, inline EDFA and Raman amplifiers are used, powered by several KV of DC supplied by the end stations. both amplifier types operate irrespective of the underlying protocol, data rate, and support multi-wavelength operation (DWDM) as well. the net result of this is that today a single fiber pair can carry up to 128 channels (lambdas) of 100Gbps traffic, or about 12.8 terabits per second. already, 200Gbps and 400Gbps end terminal equipment is in trial, and this will double or quadruple the span capacity. http://en.wikipedia.org/wiki/Forward_error_correction http://en.wikipedia.org/wiki/Raman_amplification http://en.wikipedia.org/wiki/Wavelength-division_multiplexing ar-jedi Quoted:
Quoted:
Quoted:
How deep was it? Vacuum tubes can take that kind of pressure? When they're wrapped in layer after layer of plastic, then steel, then rubber, I imagine so. Pretty damn interesting that they used vacuum tube amps, I never would have guessed that! you'll find this interesting: http://hackaday.com/2013/12/24/retrotechtacular-submarine-cable-splicing-is-serious-business/ ps regarding powered midspan amps... modern fiber optic submarine cable spans benefit from three things: increased receiver sensitivity, coding gain due to FEC (which can be quite high -- 6 to 10dB of effective gain!) , and Raman distributed amplification. using these techniques, unrepeated spans of up to 400KM are possible. for longer spans, inline EDFA and Raman amplifiers are used, powered by several KV of DC supplied by the end stations. both amplifier types operate irrespective of the underlying protocol, data rate, and support multi-wavelength operation (DWDM) as well. the net result of this is that today a single fiber pair can carry up to 128 channels (lambdas) of 100Gbps traffic, or about 12.8 terabits per second. already, 200Gbps and 400Gbps end terminal equipment is in trial, and this will double or quadruple the span capacity. http://en.wikipedia.org/wiki/Forward_error_correction http://en.wikipedia.org/wiki/Raman_amplification http://en.wikipedia.org/wiki/Wavelength-division_multiplexing ar-jedi Very interesting, I've witnessed similar splicing done in a trench on land, but with modern materials. I watched a special about the cable laying ships that can pull cables off of the ocean floor and splice them while hovering over the site. They were splicing fiber optic cables using the usual high tech wizardry, microscopes, lasers, etc. Then they built them back up one layer at a time and laid them back in place. Have you read "Blind man's bluff". Apparently we have done some tricky cable listening with subs for a long time.  Apparently it's an ongoing mission:
http://en.wikipedia.org/wiki/USS_Jimmy_Carter_(SSN-23) |
|
Quoted:
you'll find this interesting: http://hackaday.com/2013/12/24/retrotechtacular-submarine-cable-splicing-is-serious-business/ ar-jedi Damn, not even a simple leather glove when pouring lead! Forget reproductive issues, you'd be lucky if you just lost a fingerprint. |
|
Quoted:
Have you read "Blind man's bluff". Apparently we have done some tricky cable listening with subs for a long time. Apparently it's an ongoing mission:
http://en.wikipedia.org/wiki/USS_Jimmy_Carter_(SSN-23) it's all but impossible now. this is the second most-discussed topic at the lunch table. 
a DWDM (dense wave division multiplexing) system which uses coherent modulation and flat gain amps is very, very cumbersome to surreptitiously intercept. lets assume for the moment that you have devised a method of access to the submarine fiber span itself, and there is a means of optically demultiplexing the DWDM signal. the whole span, end to end, is gain balanced across the active lambdas and for the most part it would be impossible to peel off enough power of one wavelength without 1) causing lambda threshold crossing alarms (TCAs) as a result of gain tilt (this is, incidentally, how channel laser end-of-life is tracked as well), and 2) introducing upstream and downstream errors due to the coherent modulation scheme. finally, fiber optic network operators -- including submarine cable operator -- employ automated supervisory optical TDR's (OTDR) which work exactly like their electrical TDR equivalents. (aside: it is not difficult to determine, once some historical data is in place, what the average temperature of the span is using an OTDR, and track it). for example: water intrusion causes marked linearity issues in both C- and L-band transmission systems, leading to service interruptions; ergo, those operators who spent hundreds of millions of dollars dropping a submarine cable pay attention to all the little details about the submarine plant performance. it would be a really neat trick for a 3rd party to steal photons from midspan and not have anyone on the network operations side be immediately aware. as i have written before in GD, "tapping" a high bandwith facility cable is not really a good plan; on a single 100Gbps wavelength, 10GB goes by EVERY second. assuming you could write to the media that fast, a 1TB drive lasts 100 seconds. so you need 864 1TB drives, per day. every day. ok, you get 4TB drives and cut that by a factor of 4. you are making progress. did i mention that there are up to 128 "channels" (individual wavelengths) on that single fiber, each running at up to 100Gbps? since storing the data is impractical, one might suggest "scanning" it for keywords as it goes by. i encourage you to develop a 100GbE/TCPIP flow-aware pattern matcher... the folks who build backbone routers have a hard enough time routing packets at a line rate of 100Gbps, and that routing process is only looking at the fixed size header bytes, not a variable size payload. compute how deep your memory buffer needs to be if you assume each TCP packet in a given flow is separated by tens of milliseconds. for the reasons enumerated above, intercept activities are most productive and most deterministic at the client machine -- using key loggers, hacked BIOS, TEMPEST approaches, "custom" residential router firmware, and so on. ar-jedi |
|
Quoted:...the net result of this is that today a single fiber pair can carry up to 128 channels (lambdas) of 100Gbps traffic, or about 12.8 terabits per second. already, 200Gbps and 400Gbps end terminal equipment is in trial, and this will double or quadruple the span capacity.... Yep....What's really interesting, is the equipment that is attached to the ends of these fibers
|
|
Quoted:
it's all but impossible now. this is the second most-discussed topic at the lunch table.
a DWDM (dense wave division multiplexing) system which uses coherent modulation and flat gain amps is very, very cumbersome to surreptitiously intercept. lets assume for the moment that you have devised a method of access to the submarine fiber span itself, and there is a means of optically demultiplexing the DWDM signal. the whole span, end to end, is gain balanced across the active lambdas and for the most part it would be impossible to peel off enough power of one wavelength without 1) causing lambda threshold crossing alarms (TCAs) as a result of gain tilt (this is, incidentally, how channel laser end-of-life is tracked as well), and 2) introducing upstream and downstream errors due to the coherent modulation scheme. finally, fiber optic network operators -- including submarine cable operator -- employ automated supervisory optical TDR's (OTDR) which work exactly like their electrical TDR equivalents. (aside: it is not difficult to determine, once some historical data is in place, what the average temperature of the span is using an OTDR, and track it). for example: water intrusion causes marked linearity issues in both C- and L-band transmission systems, leading to service interruptions; ergo, those operators who spent hundreds of millions of dollars dropping a submarine cable pay attention to all the little details about the submarine plant performance. it would be a really neat trick for a 3rd party to steal photons from midspan and not have anyone on the network operations side be immediately aware. as i have written before in GD, "tapping" a high bandwith facility cable is not really a good plan; on a single 100Gbps wavelength, 10GB goes by EVERY second. assuming you could write to the media that fast, a 1TB drive lasts 100 seconds. so you need 864 1TB drives, per day. every day. ok, you get 4TB drives and cut that by a factor of 4. you are making progress. did i mention that there are up to 128 "channels" (individual wavelengths) on that single fiber, each running at up to 100Gbps? since storing the data is impractical, one might suggest "scanning" it for keywords as it goes by. i encourage you to develop a 100GbE/TCPIP flow-aware pattern matcher... the folks who build backbone routers have a hard enough time routing packets at a line rate of 100Gbps, and that routing process is only looking at the fixed size header bytes, not a variable size payload. compute how deep your memory buffer needs to be if you assume each TCP packet in a given flow is separated by tens of milliseconds. for the reasons enumerated above, intercept activities are most productive and most deterministic at the client machine -- using key loggers, hacked BIOS, TEMPEST approaches, "custom" residential router firmware, and so on. ar-jedi Quoted:
Quoted:
Have you read "Blind man's bluff". Apparently we have done some tricky cable listening with subs for a long time. Apparently it's an ongoing mission:
http://en.wikipedia.org/wiki/USS_Jimmy_Carter_(SSN-23) it's all but impossible now. this is the second most-discussed topic at the lunch table.
a DWDM (dense wave division multiplexing) system which uses coherent modulation and flat gain amps is very, very cumbersome to surreptitiously intercept. lets assume for the moment that you have devised a method of access to the submarine fiber span itself, and there is a means of optically demultiplexing the DWDM signal. the whole span, end to end, is gain balanced across the active lambdas and for the most part it would be impossible to peel off enough power of one wavelength without 1) causing lambda threshold crossing alarms (TCAs) as a result of gain tilt (this is, incidentally, how channel laser end-of-life is tracked as well), and 2) introducing upstream and downstream errors due to the coherent modulation scheme. finally, fiber optic network operators -- including submarine cable operator -- employ automated supervisory optical TDR's (OTDR) which work exactly like their electrical TDR equivalents. (aside: it is not difficult to determine, once some historical data is in place, what the average temperature of the span is using an OTDR, and track it). for example: water intrusion causes marked linearity issues in both C- and L-band transmission systems, leading to service interruptions; ergo, those operators who spent hundreds of millions of dollars dropping a submarine cable pay attention to all the little details about the submarine plant performance. it would be a really neat trick for a 3rd party to steal photons from midspan and not have anyone on the network operations side be immediately aware. as i have written before in GD, "tapping" a high bandwith facility cable is not really a good plan; on a single 100Gbps wavelength, 10GB goes by EVERY second. assuming you could write to the media that fast, a 1TB drive lasts 100 seconds. so you need 864 1TB drives, per day. every day. ok, you get 4TB drives and cut that by a factor of 4. you are making progress. did i mention that there are up to 128 "channels" (individual wavelengths) on that single fiber, each running at up to 100Gbps? since storing the data is impractical, one might suggest "scanning" it for keywords as it goes by. i encourage you to develop a 100GbE/TCPIP flow-aware pattern matcher... the folks who build backbone routers have a hard enough time routing packets at a line rate of 100Gbps, and that routing process is only looking at the fixed size header bytes, not a variable size payload. compute how deep your memory buffer needs to be if you assume each TCP packet in a given flow is separated by tens of milliseconds. for the reasons enumerated above, intercept activities are most productive and most deterministic at the client machine -- using key loggers, hacked BIOS, TEMPEST approaches, "custom" residential router firmware, and so on. ar-jedi Many nations we would wish to monitor don't even use fiber. Don't make the mistake of thinking every country in the world is as advanced as western European ones are...
ETA: Interesting that you use OTDRs, I am well familiar with the theory and use of TDRs in copper circuits, I've been using them for decades. |
|
Quoted:
Yep....What's really interesting, is the equipment that is attached to the ends of these fibers Quoted:
Quoted:...the net result of this is that today a single fiber pair can carry up to 128 channels (lambdas) of 100Gbps traffic, or about 12.8 terabits per second. already, 200Gbps and 400Gbps end terminal equipment is in trial, and this will double or quadruple the span capacity.... Yep....What's really interesting, is the equipment that is attached to the ends of these fibers it's keeping me employed. i used to joke that the best case scenario is that every 13 year old girl in the country has an iPhone and is watching Breaking Bad in HD... sad commentary on society, but good for folks in telecom hardware. 
ar-jedi |
|
Quoted:
it's all but impossible now. this is the second most-discussed topic at the lunch table.
a DWDM (dense wave division multiplexing) system which uses coherent modulation and flat gain amps is very, very cumbersome to surreptitiously intercept. lets assume for the moment that you have devised a method of access to the submarine fiber span itself, and there is a means of optically demultiplexing the DWDM signal. the whole span, end to end, is gain balanced across the active lambdas and for the most part it would be impossible to peel off enough power of one wavelength without 1) causing lambda threshold crossing alarms (TCAs) as a result of gain tilt (this is, incidentally, how channel laser end-of-life is tracked as well), and 2) introducing upstream and downstream errors due to the coherent modulation scheme. ar-jedi Quoted:
Quoted:
Have you read "Blind man's bluff". Apparently we have done some tricky cable listening with subs for a long time. Apparently it's an ongoing mission:
http://en.wikipedia.org/wiki/USS_Jimmy_Carter_(SSN-23) it's all but impossible now. this is the second most-discussed topic at the lunch table.
a DWDM (dense wave division multiplexing) system which uses coherent modulation and flat gain amps is very, very cumbersome to surreptitiously intercept. lets assume for the moment that you have devised a method of access to the submarine fiber span itself, and there is a means of optically demultiplexing the DWDM signal. the whole span, end to end, is gain balanced across the active lambdas and for the most part it would be impossible to peel off enough power of one wavelength without 1) causing lambda threshold crossing alarms (TCAs) as a result of gain tilt (this is, incidentally, how channel laser end-of-life is tracked as well), and 2) introducing upstream and downstream errors due to the coherent modulation scheme. ar-jedi Yep, sounds like it would be a lot easier just to hit some guy in the head with a wrench and take or install whatever you want once the lines get back on land. |
|
Damn, no phone line to, over, around, nor under the poles. |