So yesterday afternoon I was three paragraphs into an invoice email in Gmail, hit send, and got... nothing. No error, no bounce, just a blank little spinner that spun for way longer than a spinner should. Then YouTube wouldn't load in the next tab. Then Drive told me it couldn't find my files, which is a genuinely terrifying sentence to see on a Monday. Turns out it wasn't just me — Google's entire login system fell over for something like 45 minutes, and it took basically everything with it: Gmail, Docs, Calendar, YouTube, even people's Nest thermostats and smart locks, because apparently your front door needs to authenticate with a server in Iowa before it'll let you in. Google's own explanation, once they got around to posting one, was that an internal storage system ran out of space for the accounts and authentication service. Ran out of space. The company that sells you 2TB of cloud storage for $9.99 a month couldn't keep its own login servers provisioned.
I don't want to be too smug about it because obviously my own backups have failed me before (RIP the external drive that died in 2017 with three years of photos on it, never forget), but there's something almost funny about watching a company built entirely on "we will never lose your data, we have redundancy on redundancy" get taken down by what sounds like a full disk. For 45 minutes yesterday, an enormous chunk of the internet's daily infrastructure just wasn't there. Not slow. Not degraded. Gone. And it wasn't some sketchy startup, it was Google, the company most of us default to thinking of as basically weather-proof.
What got me thinking about it more than the outage itself is how much stuff quietly depends on being logged into one account. I use Google to log into my domain registrar. I use it for two different work tools. My thermostat schedule, as mentioned, runs through it. None of that is unusual anymore, everyone's account graph looks like this now, but you don't really feel the shape of it until the center of the graph goes dark and suddenly six unrelated apps on your phone are all showing the same generic "can't connect" screen at once. It's the kind of failure that's invisible until it happens and then feels obvious in hindsight, like of course this was possible, why did I think otherwise.
Anyway. Not really a complaint post, more just a "huh, that was weird" post. Everything's back now, my invoice sent eventually, the world kept turning.
Separately and unrelatedly: I finally got around to installing Cyberpunk 2077 last night after it launched Tuesday, and folks, it is rough. I'm on a PS4 (not even the Pro, just the regular one from 2015), and the game chugs like it's personally offended I'm trying to run it. Textures pop in a full two seconds after I've already walked past them. I had one mission where an NPC's face just wasn't rendered at all, this smooth blank void talking to me about a heist. I don't say this to pile on (I know half the internet is already doing that job for me), but I'd been half-ignoring the pre-release warnings because I figured they were the usual overblown hype-cycle stuff, and no, they were right, this thing needed another six months in the oven on last-gen consoles specifically. The PC version I've seen streamed looks genuinely great. Mine looks like a tech demo that escaped early.
Two very different kinds of "the infrastructure isn't ready" this week, one from a trillion-dollar company that's had two decades to get its login system right, one from a studio that promised a next-gen open world and shipped it on hardware that clearly couldn't hold it. Neither one is really a surprise if you've been paying attention to how software gets built and shipped lately — everything's bigger, more interconnected, more dependent on a hundred invisible layers underneath it, and every so often one of those layers just isn't there when you need it. Some weeks that's an inconvenience for 45 minutes. Some weeks it's a $60 game with a broken face.
I'm going to go see if there's a day-one patch yet. There's always a day-one patch.