I am remiss in posting a link to Google's peek under the kimono for usage of its free Mountain View network: The network that Google was able to build, a little late, but generally to good reviews, sees a lot of usage, the company's "alternative access" team posted to the Google blog last week. (Yes, I'm late to the game on this one.)
They have over 400 routers covering 12 square miles, which conforms to the general expectation these days of needing about 35 to 40 routers per square mile. They see 15,000 unique users each month, although we don't know with their methodology whether that's uniquely created accounts, since you can create disposable accounts on their system. They transfer 300 GB of data each day to 100 different Wi-Fi devices. Given the different number of adapters, that's the iPhone, 95 laptop cards, and four other devices (I'm joking).
Also interesting is that 95 percent of their mesh nodes see some use each day, showing that it's not just concentrated usage. I'd love to see a histogram of usage, showing the amount of usage against the percentage of routers.
Will Google build more Wi-Fi networks? No. But I can see them putting money in different forms into new models of city-wide Wi-Fi that are already underway.
"..which conforms to the general expectation these days of needing about 35 to 40 routers per square mile" This is Tropos Speak and a major spin on a limited technology and I think you really know this.
10-15 Nodes per square mile is possible when one uses a solid carrier grade Mesh Node like Strix OWS (4 or 6 Node Mesh). Even Cisco is spinning this by telling us that the more nodes you put in to cover a market the better. Their (Cisco)spin will change when they release their new 3/4 Radio Mesh systems in 2008.
The key to a good Mesh system is the throughput possible in the backhaul between nodes. In effect not having to install a Motoroal Canopy system every 3rd or 4th node to provide more bandwidth, which is how these 1 and even 2 radio Mesh systems are operating today.
Sorry, really getting frustrated by the media coverage of this (Earthlink/Tropos)Wireless Mesh Bubble Bursting and how the Mesh market is collapsing.
Jim
[Editor's Note: Tropos doesn't put out that number; that's the number coming from the field from all the deployments that are ongoing across many vendors. In Tempe, the Strix deployment was up to 600 nodes at last count LAST YEAR, and it's unclear how well covered that town is, nor whether they increased the count further.
I know you build these systems, Jim, but the physics don't work in your explanation. It's not about backhaul, it's about covering an area with Wi-Fi signals at a level that can be received with a booster indoors in an urban environment.
I should ask if you have stock in Strix? You post about them quite a lot, and I do like disclosure. I own no stock in any company I write about in this space. (I do have a few shares in JiWire, which I disclose when I write about them, and which isn't involved in hardware or deployments.)
Please point us to a functioning network that provide indoor coverage with 10 to 15 nodes per square mile--and the independent test results that confirm this. If you're talking about outdoor deployment, sorry, I don't buy it, either. If you can show a working setup and get it tested, I'll be convinced. --gf]
Hi Glen,
The "100 different devices" would really need some further explanation from Google to understand the meaning...
Cheers,
Martin