Facebook is increasingly wielding influence on the infrastructure design that stretches from the fiber optic cables to the datacenter to the last mile of your home. And it's using an open source approach to rally a vast ecosystem that'll follow its lead.
In the future, it's highly likely that Facebook will influence the behind-the-scenes network designs as well as datacenter architecture that'll run your business. And, presumably, Facebook's open source and white-box approach will drive costs down.
The progress of the Open Compute Project is fairly well known. Facebook has outlined how it approaches its datacenters and designs its servers and infrastructure. Sure, these designs revolve around Facebook's use cases, but they increasingly apply to more enterprises that have to operate at web scale.
And now Facebook is taking its Open Compute Project mojo into the Telecom Infrastructure Project. The Open Compute Project took years to land key cloud partners such as Microsoft and Google. The Telecom Infra Project has hit the ground running with partners such as Accenture, Hewlett Packard Enterprise, Broadcom, Cisco, Juniper, and a host of others.
- Facebook contributes telecom designs, plans to Telecom Infra Project, adds partners
- Open Compute Project: Gauging its influence in datacenter, cloud computing infrastructure
- Facebook wants to do to 5G what it has already done to the datacenter
- Open Compute Project enlists AT&T, Verizon while growing telco support
- Microsoft to release version 2 of its Open Compute Project server design
- Microsoft submits new open-sourced networking components to Open Compute Project
- Google joins Open Compute Project, contributes 48V rack
- Google, Facebook pause rivalries: Here's their 48V power-saving rack spec for Open Compute Project
Add it up and it's clear Facebook's approach has given it a lot of voice in how cloud datacenters are designed. It's also worth noting that Facebook's capital spending relative to hyperscale cloud providers is sizable, but the social network's influence is still punching above its wallet (not that $1 billion or so in capital expenses a quarter is chump change).
According to Stifel, Facebook will spend $4.5 billion on capital expenditures in 2016, up from $2.5 billion in 2015.
TechRepublic: Enterprise IT's top 4 favorite tech companies | Switching clouds: What Spotify learned when it swapped AWS for Google's cloud | Photos: A look inside Google's, Microsoft's and Facebook's datacenters
So how did Facebook grow this influence? Here are a few reasons:
- Facebook has leveraged open source well largely because it is a customer and not a vendor. Facebook has a core competency in running its own infrastructure, but isn't trying to sell you a cloud service or hardware. Facebook is looking to squeeze costs via commodity hardware and automation just like enterprises are. Facebook and enterprise customer interests are aligned.
- The inclination to throw designs, intellectual property, and architectural models in open source projects have won Facebook a lot of credibility in datacenter circles.
- Facebook spends a ton on infrastructure. Why would a vendor play along with Facebook? Perhaps a technical partner will sell Facebook servers, routers, and switches under a white box arrangement. Even with squeezed margins, the volume may make up the profit difference with Facebook as a customer. The other reason vendors play well with Facebook: they can learn new approaches from the ecosystem and then add their value-added intellectual property to it.
Now we all know what's in it for Facebook. By collaborating with the major datacenter and telecom players, Facebook can influence the next-generation infrastructure to enable everything from artificial intelligence and machine learning (yes, the company open sources a lot of that too) to Oculus virtual reality advertising.
However, the argument is easy to make that Facebook's approach to influencing enterprise infrastructure will have far more pros than cons.