Google Starts Running Fiber In Kansas City 118
New submitter Kiyyik writes "After weeks of wrangling over shared space on utility poles, Google and the KC Board of Public Utilities have gotten their act together and Google is starting to wire Kansas City, Kansas today. They will be paying attachment fees and hanging the fiber optic lines in the space on the poles reserved for telecommunications. The Kansas City, Missouri side is still on track to begin a few months behind the Kansas side."
Re:I can't wait (Score:5, Interesting)
It's going on existing utility poles, which tells me two things:
1) It's much, much cheaper for the initial implementation as well as any additions or repairs later on compared to burying it
2) The poles already exist, so tornadoes are likely already accounted for by the existing infrastructure
Re:This is news? (Score:2, Interesting)
"Once you have fiber to the house, you can offer what ever speed you decided to offer."
Why do people keep repeating this lie? PON has been in use for about 2 decades and in that time has speed up 40x from the first production stuff to the fastest in a lab. Point to point fiber has increased 20,000x times in 4 decades based on the fastest gear I can buy over the counter in town today.
Most FTTH is some sort of passive optical which is shared with somewhere up to 4096 other customers and one strand. This is not the yellow multi-mode fiber pairs that you can slap a 40 gigabit transceivers on and make it go faster.
Since its a single fiber, when the ONT (i.e. fiber modem) turns on its laser to talk to the head end far away, it blinds its own receiver. It also blinds the receivers for most of the other nearby ONT as well and there is a delay before they can start seeing packets again. Some companies have tried to get the transmit on one color and receiver on another but that makes things very expensive. Some places have tried optical filters with other problems but most just use the cheapest lasers they can get and live with the self-blinding problem because they are building cable TV networks.
There is also the packet coordinating problem. An ethernet packet on the 25 Gbit types PON systems is about an inch long and travels a bit faster than half the speed of light. To get 100% utilization out of your upload bandwidth, you have to coordinate the low cost optical modems to about a tenth of a nano second. The optical length of 10 km of fiber hung from poles changes by tends of meters as the wind blows it around.
PON and its derivatives are broadcast networks that started out life as a way to reduce the cost of large cable TV networks. Its not a peering network and I'm not sure it will ever be.
The network of the future will mirror the current telco networks with a pair of fibers to a central switching fabric.