allenc allencheung

A Future of Mesh Routers

Of all the new computing hardware that was announced in 2016, I was most excited to get my hands on two upgrades: the new MacBook Pro with a Touch Bar, and Google’s take on a Wifi mesh network. I’ve eyed Eero for years, but it was slightly too expensive1 to justify over singular routers, even the pricier $150–200 models that end up looking like bugs. With Google Wifi, the price of entry was only $300, and I figured that there were enough weak or dead spots away from the modem in the office that it was worth a shot.

Having had the system set up for a couple of months now, the biggest benefit has really been stronger wifi signals overall, even though physics dictates that the mesh gets progressively weaker throughput with every mesh node hop. Google wifi is supposed to be smart about handing off devices between its nodes, and includes an app that can test the health of the network plus restrict bandwidth, but these features have not been particularly useful for me.

Routers are pieces of network infrastructure, so aesthetics and features aside, their primary metric should be reliability. I had bought Apple Airport Extremes for a couple of years2, not because they looked nicer than most routers or were simple to set up with other Apple hardware, but because those routers were legitimately stable and could last years before needing a reboot.

To those standards, it was not encouraging to have my Google Wifi reset itself remotely via some Google account issue. It’s scary enough that service failures—rare as they are for Google—can cripple a local network; it also reinforced the one reservation I had about Google Wifi, in that in all likelihood, the routers will require Google’s servers to continue operating. Google has not had a great track record in keeping around projects that it deems too small to matter to their business, and they have known no qualms in bricking hardware by shutting down its complimentary services.

The trend of mesh routers backed by online services seem to only be getting started, however. There are a handful of router-focused startups now, looking to differentiate themselves on either the mesh aspect of the offering or the feature set. Again, for most folks, network features at the router level just aren’t that important; the deciding criteria will revolve around broader coverage and more reliable hardware.

And that above article on router startups actually provides an intriguing possibility of solving both issues, via embedding nodes in everyday appliances. It’s a little reminiscent of the early claims that Sony made about its Cell processor, the CPU that powered the Playstation 3 that was to be embedded all its devices and form ad-hoc mesh computing networks. It never materialized, likely due to the difficulty of programming for Cell, Sony’s inability to get its processors to other appliance/hardware makers, and the simple fact that most people don’t need the power of distributed computing in their homes3.

With appliances-as-networking-nodes, however, there is a clear utility to its users, and it solves both problems of coverage (there’s more nodes to cover nooks and crannies) and reliability (any individual node failure is unimportant). The benefits are self-evident, but the barriers to overcome will be—much like the ballyhooed Internet of Things—cost and standardization.


  1. Eero apparently agrees that their hardware could be cheaper, and recently dropped their 3-pack setup down to $400.

  2. Before they stopped making them, anyway.

  3. That said, the Folding@Home project was able to prove out that use case with a network of PS3s, so it’s definitely a matter of use case and not technological immaturity.

By allen
allenc allencheung

Elsewhere