ana_mad
Member
OP
- Joined:
- Jun 2024
- Posts:
- 298
- From:
- Madrid, ES
So I made test order with Hetzner yesterday. Their "Budget" plan looked nice—until checkout shows IPv6 as "optional protocol upgrade" for two dollars extra! Per month! Sent ticket asking why standard protocol costs more. They said "legacy infrastructure costs."
This is 2026. IPv6 is not luxury feature. I checked Contabo, same thing: IPv4 only default, IPv6 "coming soon" for premium tier only. Crazy!
Who actually includes IPv6 without nickel-and-diming? Need VPS with native /64, not NAT64 garbage. Budget is flexible but not paying protocol tax. Located in EU or US east.
Cheers!
swimming upstream since 2019 🐟
SamAlvi
Member
Self-Host Everything
- Joined:
- Jun 2024
- Posts:
- 188
- From:
- Portland, US
Get a cheap Kimsufi-alike from the scrap gods, throw Proxmox on it, one docker compose for WireGuard plus reverse proxy, you have your own /48 delegated from Hurricane Electric for free. Sure, it's a tunnel, but you control the tunnel. I run six IPv6-only containers behind nginx right now. The virtualization tax on these budget hosts is absurd anyway. Build once, scale forever.
my cloud. my rules. my 3AM alerts.
kate3
Member
- Joined:
- Jul 2024
- Posts:
- 208
- From:
- Utrecht, NL
The virtualization tax on these budget hosts is absurd
The real tax is OpenVZ masquerading as KVM. Half these "IPv6 included" providers give you a single address behind NAT with cgroup limits so tight you can't even run `ip6tables`. Ask for /64, get /128 on a shared interface. That's not virtualization tax—that's deception.
Vultr and OVHcloud are actual KVM with proper bridge networking. Verified with `systemd-detect-virt` and checking for /sys/class/net/eth0/device. Worth the small premium to avoid container-grade lies.
virsh list --all | wc -l: 47
herebungkus
Member
🌏 SEA represent
- Joined:
- Aug 2024
- Posts:
- 108
- From:
- Jakarta, Indonesia
Vultr really gives /64? Their site says "subnet" but no prefix size
wrap it, ship it, pray it
kate3
Member
- Joined:
- Jul 2024
- Posts:
- 208
- From:
- Utrecht, NL
Their site says "subnet" but no prefix size
I just spun up a $5 Vultr instance in Amsterdam to check. You get one routed /64 per location, not per instance. They delegate it via SLAAC but the prefix is static to your account+location combo. So if you deploy two VMs in Frankfurt they share the same /64. Not ideal for segmentation but it is a proper /64, not NAT64.
`ip -6 addr` shows 2001:19f0:.../64 on eth0. No NAT. No HE.net tunnel. Direct allocation.
Better than the single /128 garbage I mentioned earlier, but still not the "dedicated /64 per VM" some people assume. Their docs are here: https://docs.vultr.com
virsh list --all | wc -l: 47
ana_mad
Member
OP
- Joined:
- Jun 2024
- Posts:
- 298
- From:
- Madrid, ES
They delegate it via SLAAC but the prefix is static to your account+location combo
That is actually fine for my use case. One VM per location, separate projects. I do not need multiple VMs in same city.
Has anyone checked OVHcloud recently? Their VPS line starts at 4 EUR, IPv6 included in specs — https://www.ovhcloud.com/en/vps/. But last time I tried them the /64 took 48 hours to actually route after provisioning. Ticket dance required.
swimming upstream since 2019 🐟
