AWS introduces per-second billing and Go was started 10 years ago.

EC2 Per-Second Billing

AWS announced that they’re introducing per-second billing for EC2 Instances and EBS Volumes, which will be active from the third of October. This puts them in line with both Microsoft Azure and Google Cloud Compute, which both have had per-minute billing for a long time. And let’s be honest here, the difference between per-second and per-minute billing is small enough that it’s unlikely anyone will truly notice.

That said, it’s a welcome change as it enables more flexible tasks. Before the change it didn’t matter what you did, if you ran a 2 minute task you’d still be charged for an hour¹. Considering the hourly prices of an instance this may not sound too bad, but if this is something you do often it starts to add up. Which is especially the case if you happen to run a lot of short parallel jobs on powerful machines.

It also enables some other ways of thinking, such as truly on-demand instances. An example of this is jump boxes or bastion hosts. Instead of having a single machine that is always running, with the aid of a Lambda function you can instead spin up a temporary instance for whoever needs the access² at that very moment and have some extra security at no additional cost³.

There are still plenty of things to complain about with regards to the way AWS does its billing⁴, but this at least is a good improvement. A reminder though, if you used to take advantage of that hour by starting up your development instances, or scaling up your production machines, at 8:01, you may want to see if you should update your automation for that.

10 Years of Go

The original creators of Go wrote an article looking back on the language and how it evolved. This was done to commemorate the fact that they started work on it 10 years ago, although the official “birthday” wasn’t until it went open source 2 years later. It’s an interesting read, and I always like reading about how a language has evolved.