> thousands of monitoring checks per minute

That isn’t a lot. You could easily run that from one host. The reason people reach for Kubernetes (and similar) is because they need to scale past that single host dependency.

▲

nullpoint420 6 hours ago | parent | next [-]

100%. And a shared mental model. I love how I can scale up all my services the same way, across clouds.

It's great.

▲

tbrownaw 6 hours ago | parent | prev | next [-]

> The reason people reach for Kubernetes (and similar) is because they need to scale past that single host dependency.

I have some stuff on single-node k3s. Because it's standard so I don't have to care.

▲

ghusto 6 hours ago | parent | prev | next [-]

You could, but they don't, meaning their argument is still sound (whether they _could_ use a single host is besides the point, they're not doing that).

	▲	hnlmorg 5 hours ago \| parent [-]
		Are you sure they’re not? They’re multi-region, but that doesn’t mean they’re running across multiple hosts in each region. Docker compose doesn’t support pooling multiple hosts, so if they are running multiple hosts per region then there’s a lot more complexity to their setup than they’re documenting in that blog. Even if that complexity is human toil managing each host as a separate entity.

▲

chmod775 6 hours ago | parent | prev | next [-]

The reason most people reach for Kubernetes is because it's cool. The entire infra the vast majority of Kubernetes users have could run on a single bare metal machine with a second one for redundancy.

To be fair: using Kubernetes anyways builds the skill just in case you become one of the 0.1% who actually need it down the line.

▲

hadlock 6 hours ago | parent | next [-]

You can hire an Azure or Google Kubernetes devops guy and he will be equally comfortable on your AWS EKS kubernetes cluster. And when he leaves, you don't have a six week onboarding process with the new guy to learn all the ins and outs of your totally bespoke, non-standard container orchestration system that was cobbled together by two devs with no operations experience.

K3S takes about 5 minutes to setup the first time and you instantly have an entire universe of standardized operational tooling. I wouldn't touch docker compose with a 20 foot pole for production work.

	▲	ghusto 6 hours ago \| parent [-]
		Docker compose is hardly "totally bespoke". Setting up K8s isn't rocket science, but maintaining it are offputting, to say the least.

▲

FearNotDaniel 6 hours ago | parent | prev | next [-]

As soon as you work in a team, it’s irrelevant whether the project actually needs it. There will be someone who convinces stakeholders that it is necessary and then you just have to fall in line and learn the skills knowing that it is most likely one of the 99.9% of projects where it is just overkill.

	▲	switchbak 5 hours ago \| parent [-]
		Until your project has some success, and it turns out all those "complex" features actually turn out to be extremely useful. Which is exactly what is happening with us, too bad we didn't choose K8S from the get-go and stuck with a "simpler" tool (gaining very little in the process).

▲

nullpoint420 5 hours ago | parent | prev | next [-]

Okay, I'll bite. What if your workload genuinely doesn't fit on one machine? Like load balancing or clustering 20+ nodes for LLM inference?

▲

temp_praneshp 6 hours ago | parent | prev [-]

> The reason most people reach for Kubernetes is because it's cool.

This shittake was probably valid 10y ago, I would have agreed with you back then

> The entire infra the vast majority of Kubernetes users have could run on a single bare metal machine

Where are you pulling this out of? A large number of k8s users don't need it, but the alternative you have sounds hyperbolic.

▲

throwaway7783 6 hours ago | parent | prev [-]

..and HA