« More on small UAVs | Main | Unmet expectations »

November 23, 2006

Comments

James Bowery

Just a word of caution.

When I put up the first municipal wifi system in the US back in 2002 (funded out of hotel taxes to attract tourists to the Columbia Gorge area) the mesh boxes (running Locustworld software) we used were barely functional and required constant babysitting. The boxes were only a couple of hundred dollars then and are double that now. That's what happens with beta mesh software. You really need to be able to invest time and energy way beyond the hardware costs.

The fundamental problem comes down to the failure to follow what I called "the primary discipline" of network architecture when I was in charge of architecture at the Viewtron project back in 1981:

The user's computer is merely the host system nearest the user.

The failure to follow the primary discipline continually results in nighmarish problems with networks -- problems that plague mesh networks as well as the internet in general.

Outlandish Josh

John,

Just a note; the big problem with a lot of these services is that they provide (or rather, require) a centralized management tool. There's no technical reason for this, but it creates a great choke-point for future revenue generation.

Fon is a similar effort which attracted some white-hat hacker attention:

http://stefans.datenbruch.de/lafonera/whywedidit.shtml

The most salient point:

"While investigating the sources, we found out that FON has complete root access to the device all the time, and can supply arbitrary program code to it: Therefore, a security breach on download.fon.com (where all La Foneras fetch their instructions from) would leed to a security breach on every active La Fonera device at once. Therefore, to keep the local network secure, one has to not only trust the device supplied by FON, but the security of the FON servers as well, especially download.fon.com."

This appears to also be the case with manage.meraki.com.

Sooner or later someone will meet the demand for these tools without a single/central authority. It probably means higher equipment costs as you can't sell a router for a loss/break-even at $49 and extract subsequent revenue providing management services, but it will truly drive a decentralized proliferation of small/independent service providers.

Maraki seems aimed at making it easier for people who control a big physical area -- apartment complex, shopping mall -- to easily rent wifi to their tennants, customers. These people probably don't care about the single authority vulnerability, and if the company does well it will work for this kind of market, but it's not going to make anything interesting happen, just let you pay $5 to get online in more places than ever.

John Robb

Points taken James and Josh.

The comments to this entry are closed.