Now You Can Pay for Just One Millisecond of AWS Cloud
Computing keeps getting bigger and better, but also ever smaller.
The tech industry often focuses on the next big thing — millions of processor cores, billions of IoT devices, quadrillions of bytes of data (i.e. petabytes). But there’s another side to this constantly-expanding world. Computing is getting infinitesimally smaller, too.
Amazon’s cloud computing business, AWS, offered the latest example of the downsizing of computing at its AWS re:Invent conference. The not-so-big news: Users can now pay for one millisecond of computing power in the AWS cloud.
That’s one thousandth of one second of computing. And it costs a tiny fraction of one cent.
Until now, the smallest slice of the AWS pie that customers could pay for was 100 milliseconds. By offering its service at a fraction of that, AWS is providing an incentive to its customers — typically developers and IT professionals — to fine-tune their applications so that they can benefit from savings.
What happens in a thousand of a second? The world record for the hundred-meter dash — 9.58 seconds, by Usain Bolt — is measured in hundreds of a second. Same thing with Olympic swim times. A thousandth of a second is an even faster tick of the clock.
The flap of a bee’s wing is 5 milliseconds, according to Wikipedia. That would now be cheaper on AWS if processed using a cloud service.
How it works
AWS describes the thing that happens in milliseconds as an “invocation.” Here’s the example they give: Say 100,000 users touch a mobile phone app 20 times a day for 30 days. That would be 60 million invocations.
The capability that supports these workloads is AWS Lambda, a so-called “serverless” cloud offering that is designed to let developers scale up their apps when demand is high, scale down when demand wanes, and do so without having to worry about provisioning servers.
To illustrate the potential savings, AWS crunched the numbers for a single Web app. The cost came out to $112 when using the old-style pricing (i.e. 100 milliseconds) and $40 with the new pricing (i.e. 1 millisecond). That’s a 64 percent savings in this example.
This bargain basement discounting won’t work for all cloud apps, but for those where it does apply, the savings add up.
“With this new pricing, you are going to pay less most of the time, but it’s going to be more noticeable when you have functions whose execution time is much lower than 100ms, such as low latency APIs,” AWS said in a blog post.
It’s Moore’s Law, more or less
In some respects, this is not a new phenomenon. Tech innovation is also about doing more with less. Scaling up the data and apps, while shrinking the hardware — processors, memory, storage. At AWS re:Invent, CEO Andy Jassy also announced a server that is 1/40th the size of the previous one. He described it as AWS cloud in a pizza box.
Remember Moore’s Law? Gordon Moore, co-founder of Intel, posited 50 years ago that the number of transistors on a chip would double every 18 months, which for many years was the formula driving increases in computing power.
Some experts say Moore’s Law is winding down, but cloud computing, quantum, and other advances are still fueling this oxymoron of bigger, better — and ever smaller.