Hold 4
- vCPU
- 2 × shared
- RAM
- 4 GB
- Storage
- 2 TB Enterprise HDD
- Transfer
- 20 TB
- IPv4 / IPv6
- 1 / /64 routed
Storage KVM
Terabytes, not gigabytes.
Storage-optimised servers pairing an NVMe boot volume with large enterprise HDD arrays. Priced per terabyte rather than per core, for archives, seedboxes, backups and media libraries.
Available in all 138 locations.
The Hold line pairs an NVMe boot device with RAID-6 HDD arrays, putting the budget into raw storage rather than compute. The shared AMD EPYC 7003 CPU keeps per-terabyte cost low, but the trade-off is modest core counts and RAM. Buy Hold when your workload is dominated by capacity and sequential throughput, not when you need fast random I/O or sustained CPU. It is a deliberate fit for data that must persist in bulk, not a general-purpose platform.
Seedboxes storing large torrent libraries; backup targets receiving nightly dumps; media libraries holding thousands of films; log archives ingesting high-volume syslog streams; Nextcloud instances with sizable file trees. The large HDD arrays and RAID-6 redundancy suit sequential writes and bulk reads, while the NVMe boot volume keeps the OS responsive.
Start with Hold 4 for up to a few terabytes; step up as storage fills, since upgrading disk grows online. If your CPU or RAM sustained usage exceeds the tier, that is the too-small signal, as you cannot add cores. Choose the tier that holds your data and satisfies your transfer allowance.
| Plan | vCPU | RAM | Storage | Transfer | IPv4 | Monthly |
|---|---|---|---|---|---|---|
| Hold 4 | 2 | 4 GB | 2 TB Enterprise HDD | 20 TB | 1 | $22 |
| Hold 8 | 4 | 8 GB | 4 TB Enterprise HDD | 30 TB | 1 | $39 |
| Hold 12 | 4 | 12 GB | 8 TB Enterprise HDD | 40 TB | 1 | $72 |
| Hold 16 | 6 | 16 GB | 16 TB Enterprise HDD | Unmetered | 1 | $138 |
| Included | What it means |
|---|---|
| Full root / Administrator access | Including custom kernels, nested virtualisation and raw sockets. |
| True KVM virtualisation | Not a container. Your own kernel, your own /proc, no shared namespace. |
| Custom ISO upload | Boot anything: OpenBSD, NixOS, Whonix, a hand-rolled image. |
| Out-of-band VNC console | Recover a broken network config without opening a ticket. |
| Routed IPv6 /64 | Not a single address — a whole subnet, at no cost. |
| Always-on DDoS mitigation | Up to 12 Tbps of edge scrubbing capacity, included at every tier. |
| Three free snapshots | Instant, and taken without pausing the instance. |
| Self-service reverse DNS | Set PTR records from the panel; essential for mail. |
| Full REST API and Terraform provider | Everything the panel does, scriptable. |
| No identity verification, ever | An email address you control is the only account identifier we hold. |
Storage-optimised servers pairing an NVMe boot volume with large enterprise HDD arrays. Priced per terabyte rather than per core, for archives, seedboxes, backups and media libraries.
No — this line uses shared cores, which is why it is priced as it is. That is genuinely fine for workloads that are idle most of the time. If tail latency matters, choose Cutter or Clipper.
CPU and RAM increase with a short reboot, and disk grows online. Moving down a tier requires a rebuild, so start smaller than you think you need.
Disk grows online, and CPU and RAM increase with a short reboot on any plan. Adding extra NVMe or RAM as an optional extra is hot-added without reprovisioning where the host allows it. Moving to a smaller tier requires a rebuild, so size upward carefully.
The Hold line allocates most of its cost to large RAID-6 HDD arrays rather than cores or RAM, with a shared EPYC CPU. It is priced per terabyte, so you pay for capacity, not compute. That suits bulk storage workloads but not compute-heavy ones.