Amazon tries to sell distributed-not-decentralized computing

I initially thought that my overview of the AWS re:Invent conference would focus on the details of a bunch of new features and services. But on Thursday morning, Amazon CTO Werner Vogels took the stage and proceeded to outline a future of computing that’s utterly distributed and tailor-made for the internet of things, yet still leaves Amazon in control.

Vogels in his talk touched on everything from AWS’s new wide area network monitoring services to IoT Greengrass, APIs — even a satellite service that pulls in data from space. Vogels called the outer reaches of this distributed computing the “rugged edge.” But whatever terminology Amazon wants to use, it’s clear the computing giant recognizes that computing happens outside of its own servers. Moreover, its executives have recognized that it needs to build services designed to help its customers bridge their local compute-ingesting sensor data back to the cloud, and do so in a manner that’s both easy and secure.

Vogels started out talking about the edge, specifically the launch of 30 new local regions for Amazon’s cloud, which helps bring it closer to end users. He also touted the success of services such as IoT Greengrass, which links edge devices back to Amazon’s core cloud.

Vogels’ point was that, with the IoT, things will be massively distributed going forward, yet they will still be centralized using Amazon’s cloud and its Identity and Access Management (IAM) service. And to prove that AWS can handle the strain of a distributed-yet-still-centralized world, Vogels dropped the fact that Amazon’s IAM handles 500 billion API calls a second!

Another example of the distributed-but-not-decentralized view was the launch of AWS Cloud WAN, which lets companies link their disparate work sites under one service and control them like a single network. This is a service designed for the hybrid work world where valuable corporate data or functions are just as likely to be on an official campus as they are in someone’s garage while they work from home.

But back to Vogels and his talk. He also did a deep dive on APIs — or application programming interfaces — that had me nodding in fervent agreement. That talk was tied to the launch of AWS Amplify Studio, a new service designed to help developers build applications with visual design tools, and then the service turns those designs into code.

Back in 2018, I explained why APIs were both a bridge and a battlefield for the IoT (and frankly, all of enterprise software) as businesses rely more on software and data to derive value. In that story, I also laid out rules for folks building APIs that are almost exactly the same as the six rules that Vogels elucidated onstage. Those rules are:

1. APIs are forever
2. Never break backward compatibility
3. Work backwards from customer use cases
4. Create APIs with explicit and well-documented failure modes
5. Create APIs that are self-describing and have a clear, specific purpose
6. Avoid leaking implementation details at all costs

With the exception of the last point about leaking implementation details, which I did not include in my 2018 story, I’ve been waiting for developers to get on board the high-quality API train. It’s no surprise that Amazon, which designs everything it builds to be used as a service, has figured out a way to make it easy for developers with the Amplify Studio product.

I’m not sure if Vogels’ distributed-not-decentralized worldview is the way to move forward with the IoT, but I did shout in agreement with Matt Coulter, a technical architect at Liberty Mutual, who joined the keynote to say, “Code is a liability, not an asset.” His focus was on creating faster code execution environments and using AWS services to do so, but it applies at so many other levels.

For example, every line of code increases the complexity of your software, the complexity of integrations, and the complexity associated with keeping that software secure and free of bugs. Going forward, there is tremendous value in cleaning up the wasteful code that is slopping around our computing systems and building cleaner code.

Indeed, if the entire world is going to rely on software, then we need to view that software as a source of potential trouble, not just a means to get to a solution.

We saw a lot of other good news coming out of AWS re:Invent, such as the new version of the Arm server chips (for an excellent take, read this story) and the launch of private 5G networks, which is a really interesting option that pits Amazon against telcos and telco gear providers. For more thoughts on the conference, check out the first 10 minutes of this week’s podcast, too.