Particularly in light of AWS CEO Andy Jassy ascending to the top role at Amazon, there are suddenly a spate of articles talking about Amazon Web Services (or “AWS” as it’s commonly known). These tend to do a fair job when it comes to explaining what the hell AWS is, but they all miss the mark in one way or another.

Given that AWS was formerly relegated to mostly-technical realms, it’s starting to pop up in mainstream media. Whether it’s booting Parler off, losing a bid to serve the Pentagon, or embroiled in some open source licensing thing nobody understands, it’s getting increasingly hard to miss mentions of it.

This post is me doing my level best to explain AWS to folks who are fortunate enough to have jobs that don’t require working with esoteric cloud computing concepts.

The origin story

This is in fact the same Amazon as The company that started off as “we’re a bookstore” is nowadays “we are the freaking everything store” company based in Seattle.

Roughly 18 years ago (ish; this is ancient history in computing terms), Amazon had the revelation that if every product team spoke to one another via computer protocols instead of English, it would be a massive competitive advantage. In 2006, this idea was rolled out to customers in the first iteration of a series of “web services” that upended the entire game.

Before this existed, if you wanted to put up servers (big computers) on the internet to host a website, you had to either buy a bunch of actual computers from companies like Dell or HP or else rent them by the month from some other company. It was time-consuming and made getting a website up a fairly expensive, labor-intensive process. Servers cost thousands of dollars each, the racks to hold them and the connectivity to the internet cost thousands of dollars a month, and the only other alternatives were to rent these things from a variety of providers whose reliability was perceived to be fairly dicey. Before AWS came along, running your infrastructure for anything that was able to serve significant traffic could cost millions of dollars–upfront, in cash. Nowadays, it might still be a million dollars but it’s spread out over time as operating expenses, and you have much more control over influencing how much you spend.

AWS’s entire aim was to make the services that had previously only been available to large companies that had already made that investment available to basically anyone with a credit card. This started off with basic services such as “virtual computers,” “a storage service that would store data and return it on request,” and a couple of others that aren’t particularly germane to this explanation.

In time, these services multiplied to the roughly 200 that are available today in 2021, because they were transformative to AWS’s customers.

Roughly five years after AWS popped up, other companies caught on to what Amazon was doing and scrambled to do similar things themselves. But AWS had a massive head start.

The present day

Today, AWS is a behemoth. Netflix runs on top of it, as does Capital One, FINRA, DoorDash, and many more. When they went public, Lyft disclosed that they had committed at least $100 million a year in spend on AWS. Add them all together, and AWS currently brings in over $51 billion a year and climbing (year over year, AWS continues to be over 25% bigger!), while continuing to grow and grow.

No matter who you are, that’s serious money. But the same services that these companies are using are also available to anyone with a credit card.

One of my test accounts is free because I fit within a fairly limited “free tier” that AWS offers; the account that powers my email newsletter costs roughly $7 a month or so.

AWS has basically succeeded; there are now multiple options for anyone with an idea for a website or mobile app to spin up the computing infrastructure for relatively small amounts of money. And that infrastructure can grow with them to world-spanning scale.

The only effective limit is your budget (there’s a reason that my consulting company focuses exclusively on fixing the horrifying AWS bill), because you will absolutely run out of money before AWS runs out of computers to throw at you. For all practical purposes, AWS scales infinitely.

Amazon’s culture

The same culture that turned Amazon from a garage-based startup into the most valuable company in the world that doesn’t make laptops with insultingly bad keyboards or is run by someone who’s got a guy with a bonesaw on retainer is still very much in play. They mean it when they claim to be “customer obsessed,” that shows in the computing services they release.

Unlike Google that deprecates services seemingly for fun, AWS considers a service to be effectively a promise to its customers. They almost never deprecate existing services (their official blog has a whopping two items in the “Deprecation” category, and both were pushed back and altered so as not to break existing customer workloads). If a customer builds something on top of AWS, they need not fear that they’ll be forced to move off of whatever services they use, provided they follow the law and don’t violate AWS’s standard Acceptable Use Policy.

Of course, there’s two sides to every story, and the other side of this one is this: By never deprecating services, AWS is now a veritable sea of different services that do all kinds of things, ranging from talking to satellites in orbit to exploring quantum computing to selling machine-learning powered race cars. As a result, the services list is completely overwhelming to folks new to AWS’s very vast ecosystem; the fault for this lies squarely with Amazon. It’s also a source of growing concern for basically every company that makes software: is AWS going to release a new service tomorrow that stands to eat their lunch?

To be very clear: I am not suggesting that they slow down the pace of innovation. No customer with a painful problem is ever going to request that a vendor innovate more slowly.

Instead, what needs to change is their messaging around these services and how they align them to be a lot more discoverable by the very people each service was built to help.

How does AWS do this?

AWS is made possible by Amazon’s focus on small service teams operating nearly autonomously. As a result, new service launches and updates to enhance the functionality of their existing services become more and more frequent with time; my newsletter was my attempt to extract some signal from all of the noise.

When you consider Amazon through the lens of corporate structures, you may as well come to terms with the fact that—as compared to virtually any other company on the planet—Amazon may as well be an alien organism.

Product owners are granted a surprising degree of autonomy. And while they’re held to account remarkably firmly for service failures, unmet customer needs, and business goals, they’re not micro-managed.

This governance style is hard to replicate and definitely not a fit for all employees; if you read this and think wow, that sounds like a terrible place to work, that’s okay. In many respects, I think the same thing.

So there you have it: the basics of the AWS story.

If you’re interested in learning more about AWS and keeping tabs on what’s going on in the company, sign up for my newsletter and yell at me on Twitter.