Blog Post

Google joins Verizon in researching 10 Gbps fiber connections

The fiber is barely in the ground and already Google is thinking about next generation gigabit networks, with the advertising company researching ways to get to 10Gbps on its Google Fiber networks. Its CFO Patrick Pichette spoke at an investor conference this week and noted the efforts while telling people to stay tuned about future network expansions. google’s need for speed isn’t exclusive to it. Verizon has been testing 10 Gbps using XG PON technology since 2010.

7 Responses to “Google joins Verizon in researching 10 Gbps fiber connections”

  1. Alex, optical signals degrade over distance, which is why longer haul connections are much more complicated. We have 100 Gigabit connections, even over long haul, but as you say the electronics are pricey. The research here is bringing lower-power, kilometer-level 10 Gbps connectivity at a reasonable price for muni deployment. It’s not just a matter of taking a 10 Gbps port at each end and splicing together the length of fiber you need.

    • I must have been dreaming when I brought up NYC to DC 10G links by lighting up fiber with 10G long range optics and switches bought on ebay, a couple of optical signal boosters and a couple of regeneration stations that have switches bought on ebay. All of that for a few thousand dollars. My users must also be having some collective delusions when they see video delivered to them over those links. That’s several hundred times the distance of that a muni fiber run is going to be.

      Here’s the reality – telcos and cable cos have infrastructure everywhere – their fiber runs would pretty never be over a mile or two before hitting an active ( not passive box ) which has power. The cost of terminating 10G connection and aggregating 10G connections on the neighborhood level is microscopic if one accounts for depreciation ( which of course they do ) which means that every single thing that is being said about “cost difficulties” of the last mile is a lie. It is too bad that so called “tech journalists” actually know jack about technologies so they buy into these tales.

      Here’s what actually prevents one from getting fat pipes – it is a cost of -transit-. Even at the MFN prices when buying hundreds of gigabits of capacity with the hundred of gigabits of commits one can still expect to pay quite a bit more than a few cents per megabit per second.

  2. Oh rubbish. If you have fiber, you can make it work. If you don’t have fiber nothing that you do is going to work over the long term so you still have lay down the fiber.. We were doing DWDM with 64 colors in the nineties – it is just the gear was expensive and now it is cheap.

  3. Arty Wright

    That gear is for 10G over short runs. The long haul fiber easily handles 10G and more, but to make a single long distance thread at 10G requires different tech and lots of fiber. Long haul trunks exist in plenty, but the regional/local fibre links are patchy. I believe that’s where the Verizons and Googles are working now. Later, the issue will be providing trunks to carry the demand for very high data rate interconnect, but right now the issue is connecting everything up. PS: Don’t stop at 10GB. The clouds and supercomps need to talk at least a TB and even more.

    • Arty has it right. 10 Gbps connections in the data center are well established at Google. We’re talking about commercial fiber for municipalities, where 10 Gbps is still a work in progress — although as I tried to make clear, it’s not a work in progress that Google has a lock on.

  4. Wow. That’s impressive. Must be something amazing.

    Oh wait, how can it be that fleabay has this stuff any day for under $100?

    Any chance you folks can start asking pointed questions such as “If a 24 port 10G switch can be had for 5k in a quantify of 1 and a 10G LR SPF+ can be had for $80 in a quanity of 1 why are we pretending that there’s something special about 10G fabric?”