Our Infrastructure

Two high-performance physical nodes in Paris, France, delivering minimal latency and maximum availability.

🇫🇷

Paris — PAR1

vps-3146 · 5.180.34.45
Online
Processor
Model Intel Xeon Gold 6138
Clock speed 2.00 GHz (boost 3.70 GHz)
Sockets / Cores / Threads 10 c / 20 vCPU
L3 Cache 55 MiB (27.5 MiB / socket)
Memory
Total RAM 48 GiB DDR4 ECC
Swap Not configured
Storage
SSD
512 GB SSD
Total raw 512 GB
System
OS Debian 12 (Bookworm)
Kernel 6.17.2-1-pve (Proxmox VE)
Virtualization Proxmox VE + Docker
🇫🇷

Paris — PAR2

vps-8670 · 5.180.34.59
Online
Processor
Model Intel Xeon Gold 6138
Clock speed 2.00 GHz (boost 3.70 GHz)
Sockets / Cores / Threads 10 c / 20 vCPU
L3 Cache 55 MiB (27.5 MiB / socket)
Memory
Total RAM 48 GiB DDR4 ECC
Swap Not configured
Storage
SSD
512 GB SSD
Total raw 512 GB
System
OS Debian 13 (Trixie)
Kernel 6.17.2-1-pve (Proxmox VE)
Virtualization Proxmox VE + Docker

Where your server actually runs

Every Waifly server — free plan included — runs on one of the two nodes above, both physically located in Paris, France. You choose the node when you create a server: FR1 maps to PAR1 and FR2 to PAR2, with FR2 selected by default because FR1 reaches capacity more often. Nothing is subcontracted to a third-party cloud, and no workload is moved outside France.

What the two nodes add up to

Across both machines Waifly operates 20 physical cores and 40 threads of Intel Xeon Gold 6138, 96 GiB of DDR4 ECC memory and 1 TB of NVMe SSD. The Xeon Gold 6138 runs at 2.00 GHz with a 3.70 GHz boost and carries 55 MiB of L3 cache per socket — a 2017-generation server part, chosen for its core count and its ECC memory support rather than for single-thread speed.

ECC memory matters more than clock speed here. Error-correcting memory detects and repairs single-bit corruption in place, which is what keeps a long-running Node.js process or a MySQL database from failing in ways that look random. It is standard on server hardware and absent from most consumer machines and from a number of low-cost hosts.

How servers are isolated

Both nodes run Proxmox VE with Docker, on Debian 12 (Bookworm) for PAR1 and Debian 13 (Trixie) for PAR2. Your server is a container with reserved limits — on the free plan, 300 MB of RAM, 1 GB of disk and CPU capped at 30% — managed through a Pterodactyl panel. Containers, not virtual machines: that is why there is no root account and no SSH shell, and why you get a live console, a web file manager, SFTP and MySQL databases instead.

Availability

Measured uptime is 99.9% over the last 30 days, published live on the home page and on status.waifly.com. That figure is a rolling measurement, not a contractual guarantee — we publish what we measure. Planned maintenance is announced in advance on the Discord server.

DDoS protection is included on every plan, free one included. Waifly Guard combines network-level filtering with software protection so a server stays reachable during an attack rather than being null-routed.

Backups and retention

Backups are included on every plan: one automatic backup slot on the free, Nano and Mega plans, two on Giga. If a server is suspended, what happens to its data depends on its age — a server created less than 7 days earlier is deleted after 3 days of suspension, while an older one is retained for 15 days and can be recovered from backups.waifly.com. The anti-abuse rules set out the full thresholds.

Scale as of today

These two nodes currently carry 377 deployed servers for 1.1K registered accounts. Both counters are read live from status.waifly.com and refreshed daily in this page, so the figures above are current rather than aspirational.