An API-first KVM provider aimed squarely at engineers: blank Debian, Alma or Alpine images, per-second billing, and a documented REST API with a first-party Terraform provider. Dedicated-core and shared-core lines across 14 regions, NVMe storage throughout, and 10 Gbps ports on the larger plans. There is no managed layer at all - support will not log into your box, by design.
Share your experience to help others choose the right provider.
Running 11 nodes here for about eight months now. The Terraform provider is genuinely first-class - plan/apply does what it says, no phantom drift, and the API returns useful errors instead of a 500 and a shrug. Boxes come up from a snapshot in under 40 seconds. Support is email-only and will absolutely not touch anything inside your OS, which suits me, but know what you're buying.
No complaints about the servers themselves. But my ticket about a broken IPv6 route sat for 31 hours before a human replied, and the reply was a link to a status page that didn't mention the incident at all. If you need hand-holding, this is not your host.
Moved a Postgres read replica off a shared-vCPU host elsewhere and the difference was night and day. fio gives me a consistent ~180k random read IOPS on their NVMe pool, which matches what the plan page claims, and I've seen steal above 1% exactly twice in six months. Only real gripe is that attachable block storage is still limited to two regions.
The panel is fairly bare and I still can't reorder firewall rules without deleting and recreating them. But the documentation and CLI more than make up for it, and rescue mode plus a working serial console have saved me at 2am more than once.
If you want cPanel, one-click WordPress or someone to debug your reverse proxy, look elsewhere. If you want a clean Debian box, an API key and per-second billing, its hard to beat. I spun up and destroyed roughly 60 test instances last month and the invoice was under nine dollars.
Running 11 nodes here for about eight months now. The Terraform provider is genuinely first-class - plan/apply does what it says, no phantom drift, and the API returns useful errors instead of a 500 and a shrug. Boxes come up from a snapshot in under 40 seconds. Support is email-only and will absolutely not touch anything inside your OS, which suits me, but know what you're buying.
No complaints about the servers themselves. But my ticket about a broken IPv6 route sat for 31 hours before a human replied, and the reply was a link to a status page that didn't mention the incident at all. If you need hand-holding, this is not your host.
Moved a Postgres read replica off a shared-vCPU host elsewhere and the difference was night and day. fio gives me a consistent ~180k random read IOPS on their NVMe pool, which matches what the plan page claims, and I've seen steal above 1% exactly twice in six months. Only real gripe is that attachable block storage is still limited to two regions.
The panel is fairly bare and I still can't reorder firewall rules without deleting and recreating them. But the documentation and CLI more than make up for it, and rescue mode plus a working serial console have saved me at 2am more than once.
If you want cPanel, one-click WordPress or someone to debug your reverse proxy, look elsewhere. If you want a clean Debian box, an API key and per-second billing, its hard to beat. I spun up and destroyed roughly 60 test instances last month and the invoice was under nine dollars.
Two unplanned outages in their FRA region inside five weeks, one about 40 minutes and one closer to two hours. The postmortems were detailed and honest, which I do appreciate, but the SLA credits came nowhere near covering what the downtime cost us. Amsterdam has been rock solid by comparison, so it seems region-specific.