Published in


How AWS and Other Cloud Providers Became the Internet’s Most Powerful Moderators

In another era, Parler would have owned its servers—and remained online

Photo: Thiago Prudêncio/SOPA Images/LightRocket via Getty Images

When Amazon Web Services decided to stop hosting the alt-right social network Parler last week following the insurrection at the Capitol, it looked like the site was doomed to go offline.

Migrating an app successfully between cloud providers, and ensuring it works on the other side as expected, is hard enough. But moving the vast amounts of data associated with a social network (likely hundreds of terabytes of information) would be agonizingly slow, taking far longer than the 24-hour warning Amazon gave Parler.

Unfortunately for Parler, virtually every other vendor was ditching them as well. With cloud providers rejecting them and no physical servers of its own, Parler has nowhere to go and now says it may never return.

The swift shutdown of Parler illustrates a wonder of the modern internet. It’s simple to get a website or service online without ever physically seeing or touching a server. Developers can choose from an array of hosts, from Amazon Web Services to Microsoft Azure or Google Cloud, click a few buttons, and be online in a few minutes. These companies manage vast data centers full of servers, renting them out by the hour, so you don’t need to think about setting up your own gear. They have centralized much of the web, which gives them unprecedented power to police it.

This is a relatively new phenomenon. Until just over a decade ago, getting anything online at scale was a complicated, expensive process. It required procuring expensive physical servers from a company like Dell or HP, putting them in a data center somewhere, and configuring them yourself to get them up and running. Smaller companies might contract with a “colocation” provider like Rackspace to rent a spot for their servers in its data center, but the arrangement was expensive and slow, taking days or weeks to get the contracts signed, servers delivered, set up, and online.

Those enormous data centers are still there; they’re just operated by the largest companies at an enormous scale. Rather than renting an entire server sitting in a rack somewhere, virtualization technology allows companies like Amazon to rent out hundreds of tiny slices of a single server’s performance to lots of people, which is what makes it such a profitable business.

Because of the complexity and expertise involved, nobody wants to buy an actual server because then you’re stuck maintaining it, replacing it every few years, and making sure it’s in a safe place — doing so would be far too expensive for most services.

I know how difficult all of this is because my first job out of university, as a network engineer, was doing exactly that: managing and deploying physical servers in data centers for clients. I graduated into the dawn of the cloud age when most companies still preferred to buy their own hardware and host it on their own premises in their own special data center.

It was grueling, slow work: Unboxing giant 60-pound servers as well as large storage arrays full of hard drives, putting them into the rack, cabling them up, and configuring them before ultimately putting them online. It’s unlikely that Parler — or anyone else — could manage such a feat within a month, let alone just a few days, not to mention that doing so would be tremendously expensive.

Larger companies and governments do still operate their own data centers for various reasons, from data privacy to local laws that disallow using cloud services, but the vast majority of the world now rents servers from one or more cloud providers.

Most of the largest services, like Apple’s iCloud, use multiple cloud platforms to avoid putting their eggs in one basket in case of outages. The average company can’t afford this because it’s hilariously expensive to pay multiple cloud platforms—and it requires ensuring that your app is built to work correctly when spread across all of them.

Snap, for example, revealed in 2017 that it had a contract with Google Cloud for $2 billion over five years for its service, in addition to a contract with Amazon Web Services for $1 billion. It’s unlikely Parler has that kind of money, let alone custom contracts with any of these companies.

When Amazon booted Parler from its cloud platform, there was little the company could have done to stay online, which shows the power of deplatforming hate. A decade ago, when the norm would have been owning actual servers and space in a data center, the company likely could have stayed online in a similar fashion to how torrent website The Pirate Bay famously dodged almost every global takedown it’s faced over the last 20 years.

But, gone are the days when it was common for smaller companies to own servers. It’s simply not realistic anymore. Getting things online fast and cheap has been an incredible change to how the internet is cobbled together, but it has also centralized the majority of the web around giant platforms like Amazon Web Services, Microsoft Azure, and Google Cloud — which means that the internet is at the mercy of their whims as well, for better or worse.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store