Google Planning New Undersea Cable Across Pacific?

tregetour writes "Google is planning a multi-terabit undersea communications cable across the Pacific Ocean for launch in 2009, Communications Day reports: 'Google would not strictly confirm or deny the existence of the Unity plan today, with spokesman Barry Schnitt telling our North American correspondent Patrick Neighly that "Additional infrastructure for the Internet is good for users and there are a number of proposals to add a Pacific submarine cable. We're not commenting on any of these plans." However, Communications Day understands that Unity would see Google join with other carriers to build a new multi-terabit cable. Google would get access to a fibre pair at build cost handing it a tremendous cost advantage over rivals such as MSN and Yahoo, and also potentially enabling it to peer with Asia ISPs behind their international gateways — considerably improving the affordability of Internet services across Asia Pacific.'"

Re:evesdropping requirements

[$RANDOMLUSER]

...can splice without interruption...

For copper, sure, but not with fiber optics.

Re:evesdropping requirements

[AHuxley]

Its already set up via the NSA's Kunia Regional Security Operations Center in Hawaii.
NZ, Australia, Japan and now something extra in Hawaii. Asia is now so tapped.
Google is of no interest, the NSA can tap at any point they want.

http://cryptome.org/google/kunia-us.htm [cryptome.org]

Re:why is this better than satellite upload/downlo

[Anonymous Coward]

Satellite is great for some applications, but lousy for real-time communications.

Video surveillance and satellite would work fine together: a 1 second delay usually isn't a concern in such applications. Same thing with batch jobs and large file transfers.

But for short message/interactive applications (games, shells, telephone communications), an undersea cable is, right now, the best communications path. Very high bandwidth and shortest-path. The big downside of a cable is that it is more vulnerable to damage by nature or by vandalism.

Satellites are awesome for some applications, but they have significant trade-offs. Namely, expensive, unrepairable equipment, jamming potential, and the highest conversational delays.

Re:evesdropping requirements

[Iphtashu Fitz]

Considering that the US Navy in conjunction with the CIA was tapping Soviet copper phone cables as far back as the 1970's [divingheritage.com] I wouldn't find it all that unlikely that they now have the technology to tap fiber cables. Yes, I know that splicing into fiber is extremely difficult in the best of situations, but if braniacs could figure out how to locate and tap underwater copper cables almost 30-40 years ago then I wouldn't hold it against modern-day braniacs to figure out a way to tap fiber cables in this day and age.

They are buying one fiber pair

[mbone]

Don't get the vapors, everyone. Google is buying one fiber pair. This will lower their costs, but only that. There will be, what, 200+ fiber pairs in that cable. There will be some to go around if anyone else wants to pony up.

As for "considerably improving the affordability of Internet services across Asia Pacific,'" I don't follow that at all. Google doesn't sell transit. The new cable might do that, but not because of Google - because real ISPs will get other fiber pairs and use them to sell transit.

Next, we'll get articles about how Google's corporate jets will revolutionize air transport in North America ! (At least, for Google execs.)

Neal Stephenson article on cable laying

[Rick Bentley]

Neal Stephenson did an article for Wired on the laying of global fiber optic cable about a decade ago. It's a long read but a good one (kind of like Snow Crash was). He travels around the world following the laying of FLAG (Fiber Link Around the Globe). He covers everything from laying the cable, to the landing points, to over-land connections, to telco monopolies, to everything else. If you're a geek and into submarine cable laying, then the article below is almost required reading. http://econ161.berkeley.edu/OpEd/virtual/stephenson.html [berkeley.edu]

Re:They are buying one fiber pair

[bridson]

> There will be, what, 200+ fiber pairs in that cable.

Actually modern submarine fibre-optic cables usually contain four or less fibres. The massive traffic capacity is provided by multiplexing wavelengths down the same fibres. A modern terminal can typically handle up to 192 wavelengths @ 10Gbps (hence the multiterabit capacity).

Therefire ownership of a complete fibre pair in one of these things is a significant investment!

http://www.alcatel-lucent/submarine [www.alcatel-lucent]

Re:evesdropping requirements

[DDLKermit007]

At which point the owner notices a drop in signal quality. That video is a scare piece (hell it's a fucking marketing piece for a company who wants to sell their hardware encryption tech). Their caveat that gives it away is "unless there is equipment is in place to check for signal degradation". You flatly do NOT setup a fiber system, not to mention one as big as what Google is debating without the proper equipment. Hell, you really don't even need equipment. Just software to check for degraded data.