AWS Data Centers in Bahrain and UAE Destroyed Beyond Repair by Iran - USA Loss to IRGC Getting Worse

Eli the Computer Guy (0:00)

I believe the technical term for this is, well, damn. So coming from CNBC, AWS says it can't restore service to Bahrain UAE facilities six months after Iran strikes. So I kind of think this is interesting in the modern world when we think about IT architecture, and we think about things like resiliency, and basically what happens when people start trying to plink your data center. So one of the big problems we've had over the past decade or so is all of these companies have started moving their systems to the cloud. We're migrating to the cloud. The problem is, is the cloud is an actual thing. When we talk about the cloud, right? So back in the day, we knew exactly what physical box all of our services were running on, right? You would walk into a server room, that box runs our Active Directory server, that box runs our file server, that box runs our VPN server, that box runs our Exchange server.

Basically, the services that were being provided, they were installed on the operating system, the operating system was installed onto the hardware, and so it was all one thing. When we start talking about the cloud, what we're talking about is the modern world of systems architecture. Kind of services are able to move around on hardware. These things are called virtual machines and, oh, what are they called? Linux, Kubernetes, Docker.

Anyways, there's this thing. If you know what Docker is, you know what I'm talking about. Containers, there's containers, right? Anyways, you have virtual machines, you have containers, you have clustering and you have all this kind of stuff. And so nowadays in the modern world, even if you walk into a data center that you own, that you control, you may look at a rack of servers, and you're not really sure where the VPN server service is actually at this period of time. The idea is that you can migrate around these virtual machines or containers or whatever services to other hardware. And so the idea is that it's a cloud, it's like cloud. So a private cloud is if you have a rack of physical machines, and on that are services that are just doing whatever the hell they're doing. That is a private cloud. A public cloud is when it's up on Amazon, AWS, Microsoft Azure, DigitalOcean, that type of thing. The big thing to understand though is when we talk about the cloud, is the cloud is still servers. It's still servers, it's still data centers, it's still power distribution, it's still cooling, it's still all of that kind of stuff. You just don't physically know. Like when you spin up a VPS on, let's say DigitalOcean, or you spin it up on DigitalOcean, you get an IP address back, and somebody says, oh, where is your VPS? You say DigitalOcean.

You don't really know where it is. Like you might have a region, you might know it's somewhere in St. Louis, or somewhere in New York City, or somewhere in Los Angeles, but you don't know really. But the important thing to understand though, is that there is, there is a physical machine somewhere that is actually running the services that you are gaining access to. And so one of the things to consider in the modern world when we start thinking about natural disasters, when we start thinking about war, when we start thinking about these things, is thinking about with your architecture and how you design your architecture for resiliency, whether you have thought about the concept of what happens if somebody starts blinking the data centers, blinking the cooling towers, because that's all you have to do. To take down a data center, you don't need to burn down the data center, you don't need to blow up the data center, you don't need to go in and shoot every damn server in the data center. You can just sit back a decent way, right? What is it? Center of mass, you can just simply center of mass a cooling tower blink enough of those motherfuckers and the entire thing goes down. And when I talk about that, it's like, oh, that's Eli being fanciful.

But here in North Carolina, we had some jackasses that decided to start plinking power conduit insulators. So when you have these power lines, there are these insulators between the power line and the towers themselves. Yeah, they just started plinking those things. I mean, think about that. They were able to do a tremendous amount of damage by just plinking some insulators.

So one of the things to be considering right now is when you've designed your architecture, have you designed your architecture in such a way that if somebody starts plinking the cooling towers of the data center, your machines, your virtual machines are currently running them, what happens post that? I think about this a lot with Amazon. So Amazon has regions, right? So Amazon has numerous regions, and so when you set up your cloud services, you can set up your cloud services to replicate to different regions. Now, for some reason, again, this is just my bias, my whatever, I just noticed this, it seems like region one for AWS drops quite a bit. Seems like it drops a lot. Anyways, and so when the entire region goes down, then all of your services go down, even though it's on the cloud, right? It's within the data centers within region one. So if something happens to the data centers in region one, all of your services are down for a while. Now, one of the things that Amazon says is, yeah, shit happens, right? Part of your terms of service is once in a while, a region might go down, and that's why they've talked about, there's design specifications so that your services should replicate to multiple regions. So you have your services in region one, and region, I don't know the specific geographic regions, but let's say you have it in region one, and you have region three, and you replicate between those regions. If region one drops, all of your services are still maintained in region three. If region three drops, all of your services are still maintained in region one. One of the things that Amazon says though, is multiple regions might go down. So both region one and region three go down, then you're still down. So maybe you want to replicate to region one, region three, and region seven. Anyway, the problem is, once you start replicating, right, that's more money, more money. Like Amazon's like, look, we're not saying there's going to be a problem, but there might be a problem. So you might, you might want to pay us a little bit more money basically in protection services, all right? And so that's the thing, right? So you, so, so the more regions you replicate to, the more reliable your systems are going to be, but the more money it's going to cost. And so you go to your CEO, you go to your executive and basically say, I need more money. And the executive is going to look at you like, wow, what is this going to give us? He's going to say resiliency.