Silence because no ampere. No A100. No H100. No MI250.
Contabo price is beautiful. Beautiful like painting. You look, you smile, you cannot train.
My RTX 4090 runs 24/7. Electricity bill makes me cry. But at least CUDA works. Contabo "dedicated" means "dedicated to disappointment" for GPU.
Different thread? Yes. Same pain? Also yes.
I have 14 status pages. OVHcloud is one. Contabo is not on the list because Contabo has no status page I would trust.
847 days sounds lovely. I have 847 days of screenshots from UptimeRobot, Pingdom, StatusCake, and eleven others. The ones that matter are the ones where the provider's green checkmark and my red alert overlap. That overlap is where you live or die.
"Live-patched" is lovely until the patch fails open. Seen it twice. Not with OVHcloud, with another shop. But I saw it. I have the graphs.
Member
Difference is you also don't disclose whether that booking flow handles fifty concurrent or five thousand. "Works" is not a benchmark. "Works at what load, with what failure modes, under what recovery time" is a benchmark. I use https://github.com/masonr/yet-another-bench-script for that.
The cake story is charming. Charming does not show up in a postmortem.
My point about silent churn stands. People leave without announcing it. The ones who stay and post have reasons. Sometimes those reasons are technical. Sometimes they are social. Telling them apart takes more than uptime counters.
