OVH has been getting a lot better.

They recently discovered that Terraform exists and have a usable infra as code provider now. They're starting to take multi-AZ seriously. Sometimes their network is UP and working normally, which compares favorably with us-east-1.

It's starting to look like a real cloud.

Care to elaborate? I've been using their different products for around 20 years.

Ovh has a lot of products. Some of them are cheap and the services quality acts accordingly. OVH is mostly pretty good.

While not using it on a huge scale, I've got a few projects in OVH cloud(US) and can't remember them having ever gone down in the last couple years.

I get maintenance notices from them often that explain what they're doing and that it shouldn't be impacting, and so far they've been right.

Is your experience different?

There is also this weird question: What do you get from your hoster?

For example, I've had endless discussions with people about the reliability of Hetzner Dedicated Servers. At the end of the day, you have to realize: You get a physical server, with fans (we had performance degradations because a cable binder degraded and parts fell into a fan and the CPU throttled), a PSU, drives (HDDs and SDDs fail differently, but both fail. SDD failure can be much more evil). It's just a little box that can run for years, or it can choose to go kapeister whenever it wants. Maybe it will take it's friends along the way too. There have been outages of servers catching on fire and frying the systems on top of them as well. Then the fire suppression goes off and shatters some drives plates on top of that. Naturally only in the archiving servers, who'd be using spinning rust in other systems this day and age?

And that's the operational technique and experience that has been hoovered up by very large PaaS offerings and hosters. You need to plan for, deal with and mitigate the situation that every server and VM hosted on a server (read: all of them) are a somewhat useful crew of saboteurs that are trying to figure out when the right failure of 3-4 systems are going to cause you a lot of overtime, stress, and maybe impact and cripple the business as well.

If you plan for this, Hetzner Cloud + Dedicated can be a great hoster, with great support and really good value for money.

If you assume that a single Hetzner Dedicated Server or Cloud VM has the same manpower behind it to give it the staggering uptime of EC2 instances, and you bet all of your company and all of your money on this VM never going down... well, you can do this on AWS. We've had a prolonged outage of an EC2 instance once in like 7 years.

But don't do this. Fix your failovers and architecture and embrace the fun of european hosters. After some grief with the early stages of the Cloud-Dedicated-VSwitch infrastructure, we're seeing great uptime with them.