Hacker News Reader

Comment by kristopolous | Hacker News Reader

This sounds like a common and generic enough business requirement. Isn't there something off the shelf that they could have used instead?

View on HN

Sure, it's easy enough to find network-attached screens. The hard part is giving those screens access to anything without them showing up as vectors for every Red Team exercise.

Gizmos are used for basically any screen at Facebook that needs to display "live" information -- conference rooms, building maps, monitoring dashboards, etc. Each one has a provisioned identity protected by secure boot and hardware root of trust. They have a minimum of onboard storage (maybe even nothing mutable/persistent after boot?) so they're not a foothold for persistence.

So they're essentially Facebook's internal raspberry pi with a ton of accessories + focus on not becoming security nightmares?

There are many reasons why Facebook could prefer going with homegrown hardware:

1) It allows them to actually work with hardware and cross pollinate knowledge to their consumer hardware businesses. And I’m not talking only about the hardware itself but all the institutional knowledge required to negotiate with vendors, outsource circuit design and procure parts.

2) They can create one device that is multi-purpose like this one and control every low level detail so it can be used even for security critical tasks.

3) They can test the devices against their own infrastructure and collect low level data that is useful for developing video/voice services and encrypted services in real case scenarios.

In general it just makes sense. Facebook’s market cap is 6X larger than Cisco and it employs 50k people. At that scale and for a company like FB, this doesn’t sound crazy.

Unless I’m missing something, none of those reasons sound terribly compelling to me, given we’re talking about a social media company. Facebook is fundamentally not a hardware company, things like Oculus notwithstanding.

There's some projection-psychodrama to it, but it goes along with the tape on Zuckerberg's webcam: they really don't want Google, Amazon, or their Chinese competitors on their corporate network, especially not with cameras and microphones.

Not a hardware company yet. There was a time where Amazon and Google were not hardware companies.

But I think saying they are not a hardware company is a bit unfair. Oculus by itself is a tremendously hard and complex hardware business. Portal less so, but it’s still a meaningful device with a lot of complexity.

Also Facebook infrastructure must have hundreds of proprietary hardware elements that are built specifically for their own use cases.

But I get what you’re saying. From all the tech behemoths, FB is the by far the least experienced in hardware. But I think it’s pretty clear that FB sees hardware as a column of their long term vision. That alone is enough reason to invest on developing your own hardware.

How about just as an exercise in actually having fun? I'm sure there are several qualified hardware engineers at Facebook and the company has plenty of money. All of the off-the-shelf solutions to this problem are complete garbage, so they just decided to make a better one. For fun.

Facebook started building their own hardware prior to 2011 (when they open sourced their designs): https://about.fb.com/news/2011/04/facebook-launches-open-com...

There are literally thousands of these machines at each office.

Maintaining a fleet of android devices sucks hard, especially when they are on devices that are not really designed to be remotely controlled (any kind of tablet is really not that well designed to be managed remotely (ie has batteries, only works on wifi, not onboard TPM thats useful))

Plus they are posted in every room where sensitive conversations are going on, not only that there is a list of who's going to be in those meetings. Not only that, they have access to video conferencing in each room as well.

Combine that with the portal offering, I can see the compelling value add.

as for "not a hardware company" their datacentres are almost entirely custom.

But they are big enough that most hardware companies would be small-scale operations in comparison. Why wouldn't they set up a division to do something like this?

Not a hardware company? Haven’t they been making their own DC power servers and high end SDN gear for a half dozen years now?

Gizmo was from the group I worked with at Facebook. This was almost entirely from a "How can we make it easier for people to collaborate". 2 is the closest reason :)

They were multi functional:

- Room Check-in devices (replacing Nexus 7s)

- Building Maps (replacing mac-minis)

- Reception Check-in (replacing iPads)

- Employee Dashboards (replacing mac-minis)

- Video conference control panels (replacing the cisco device)

The advantages they had were:

- Single SKU for IT Operations

- Pulled down a fresh image on each boot and had no state which made IT and security happier

- Could join the correct VLAN via 802.11x, the key lived on the TPM chip which made security happy

- Powered via PoE allowed remote restarts

This is something companies would pay huge amounts of money for.

Any idea why Facebook never marketed these?

Presumably it was envisioned as an internal project, so nobody though about trying to sell it: pure inertia.

Maybe some version of the innovator’s dilemma? The product would have to be insanely successful for it to move the needle at such a large corporation.

> This is something companies would pay huge amounts of money for.

> Any idea why Facebook never marketed these?

This was designed to fit with internal Facebook frameworks and network capabilities. The device itself does not do much, in most cases just starts a browser and points it to an internal webapp. The device is just the presentation layer of a great deal of backend work that can't be easily extracted as a product.

Still, I'd take that over some shitty "slap a $50 android tablet on a wall and point it at our weird http site" -solution.

> The advantages they had were:

- Get "misplaced", nothing of value can be stolen.

Considering the results the twitter user had, mission accomplished.

The first thing I thought of when I saw this was this was almost certainly some sort of precursor/inspiration for the Facebook Portal which is a video conference focused smart....thing. I'd say your #1 is really on point.

> certainly some sort of precursor/inspiration for the Facebook Portal

They were developed independently.

Source?

I've been part of the team that built the device OP talks about.

In my experience, most of the off the shelf systems for stuff like this doesn't work very well. Even when it does, it's not extensible. I could see it being appealing to have a more customizable in-house system, at least when you have as much engineering talent as Facebook probably does.

Also when you want to train your in house engineering to make hardware, good to produce something internal and dog food it.

There’s a lot. Cisco makes them if you have their conference room hardware. Any new tech corporate office (and some non tech) will have some sort of hardware that lets you dial into the meeting and or reserve the conference room.

Sure, but it’s either too expensive (Cisco), too unreliable (MIMO) or both (Tandberg). Source: used to be a Service Desk tech for a giant conglomerate.

Reply on news.ycombinator.com