I’m not a backend developer but I do find the care and feeding of services fascinating.
In this video Christina Warren takes us into a GitHub lab — complete with lab coat — to discuss GitHub’s M1 Mac runners.
I’ve wondered how these machines were setup. I’d imagined a bunch of Mac Mini’s turned vertically in a custom rack, but that’s not at all what GitHub does! They disassemble the Mini, pull the parts and pieces out, and put them into a custom sled that allows them to slide right into a rack. It’s extremely cool and now I want a rack of them in my home. 🤣
This did raise some questions however.
What do they do with the shells after they gut the Mini? Do they have a deal with Apple to return them? Do they fill them with PC parts and use them as desktop computers?
Why are they using M1s? Why not M2s? Is it because it took time to build out the new sled and master gutting the Mini without destroying parts? Is it because the M1 is so much better than the old x86 boxes that they decided to get M1s at a discount?
How many total Mac runners do the currently have at GitHub, including x86 based boxes? Were they all gutted and configured to work on a sled like these new M1 versions?
Is the process of building out a sled and rack of these documented anywhere?
Does Apple have an opinion on this? Does Apple have a similar setup for their Xcode Cloud service or do they just have a room full of Mac Mini’s somewhere?
Does GitHub have more Macs on their public facing network than Apple? I’d imagine Apple is using cheap PC boxes running Linux for any service they’re not already running on AWS or Azure.
Is Apple, perhaps, running their Xcode Cloud service inside GitHubs infrastructure?
So many questions.