That's the risk transfer in a nutshell. Hosts want residential-grade pricing with industrial-grade marketing. "Cloud" implies redundancy you don't have to think about. Then the fine print says "actually you do have to think about it, and also we can turn you off."
I don't blame HostHatch specifically. I blame the industry for normalizing this.
Member
I tested this properly before I complained.
Signed up for a CloudCone $3/mo instance, set up external monitoring via UptimeRobot and my own Nagios. First 60 days: zero events. Then June heat wave, Los Angeles grid, 19-minute DR event. Nagios caught it. UptimeRobot did not—too brief for their 5-minute check interval.
So your monitoring might not even see it. Your users will.
Following this thread closely. I have a HostHatch storage VPS in Amsterdam that I use for backup. 20 minutes offline during a DR event means my nightly rsync fails and I get an email I have to read. That's it. Not the same problem as FlowSana's database cluster.
The risk transfer depends entirely on what you're doing.
Mi homelab tiene homelab and also now tiene a UPS because the Buenos Aires grid makes DR events look like a vacation
Seriously though: if you are in a place with bad power already, you engineer for interruption. If you are in a place with good power, you assume the host does that engineering for you. The host is now saying "no, we participate in making the power worse on purpose." different assumption.
Member
Exactement. I self-host at home in Lyon because Engie grid is stable and I control the UPS. My HostHatch VPS is for off-site backup. If they shed load, my backups pause. I do not pause.
But I am also not paying 60 USD for this. The price shapes the expectation.
Has anyone actually collected on the "credits offset any compute time lost"? HostHatch's credit is minute-for-minute, which for a $5/mo VPS is $0.007. My time to open the ticket and document the event is worth more than the credit.
The credit is not compensation. It is a procedural fig leaf.
Fair point on the math, and we hear this. The credit mechanism is specified in 14.3 because it's auditable and automatic, not because it fully makes you whole. For business-critical workloads, we do recommend our higher-tier plans with explicit availability commitments, or multi-region deployment.
We're also reviewing whether to offer a DR-free tier at a premium, as some of you have suggested. No timeline yet.
DR-free tier would be honest. Current setup is "same price, hidden lottery." Some customers get interrupted, some don't, depending on facility and luck. The ones who read forums know. The ones who don't are the marks.
I went and checked. HostHatch's Madrid facility is on Iberdrola grid. Spain has mandatory demand response for large consumers since 2024. So Madrid is in. Oslo is on Statnett, which has voluntary programs. So Oslo might be out, might be in, depending on whether HostHatch enrolled.
This is not information they publish per facility. You have to infer from grid operator public filings.
Member
Statnett's FFR program is voluntary but heavily subsidized. If HostHatch has a facility there, they enrolled. The economics are too good. Same for any host with meaningful load in the Nordics.
Zurich is the exception because EKZ's program structure is different and the Swiss don't subsidize it the same way.
