Offsite Backup Server Hosting — Crypto, 1–5 TB NVMe

An offsite backup target you control: root KVM with 1–5 TB NVMe, borg/restic/rsync over SSH, snapshots, no KYC, crypto-only. From $24/mo.

Published 2026-08-30 · Updated 2026-10-21 · BitCloudVPS Editorial

TL;DR — your own backup box, paid in XMR

A Storage VPS at $24/mo gives you 1 TB of NVMe as a full root server for borg, restic or rsync — with snapshots, unmetered 2.5 Gbps and a crypto-only invoice. Backup boxes are the classic offsite target: your keys, your retention, no vendor reading your files.

Why not a Storage Box

Hetzner's BX line is cheaper per raw TB (~$3.50/TB) but it is a protocol box: SMB/NFS/rsync endpoints, no root, no custom agents, and Hetzner can demand government ID. A backup target you cannot script freely is not yours. BitCloudVPS Storage is a root KVM: your scheduler, your encryption (borg/restic encrypt client-side anyway), your retention, paid in Monero if you want the payment private too.

Setup that works

Debian + borgmatic or restic over SSH; SSH keys only, password auth off. Client-side encryption before anything leaves your network. Snapshots add-on ($9.50/mo) for point-in-time recovery. Restore drill once a quarter — a backup you never restored is a hope, not a backup. 3-2-1 still applies: this box is the offsite copy, not the only copy.

Related

Compare: best storage VPS 2026. Payment: buy storage with crypto. Bigger disks: Storage 1–5 TB plans. Mail/Nextcloud angle: self-hosted mail.

What a year of offsite backups actually costs

Price the backup box by the year, not the month, because backups only prove themselves over long horizons. The Nano plan lists at $5.49/mo, but the flat annual price is $36 — an effective $3/mo — which is enough for encrypted restic or borg repos of a personal workstation. The Starter plan at $8.50/mo adds a second EPYC 7443P vCPU, 4 GB ECC RAM and 60 GB NVMe, which suits a small office's database dumps plus document trees.

When the dataset outgrows that, the storage line runs from $24 to $56/mo with 1-5 TB of NVMe and up to 8 vCPU and 32 GB RAM at the top tier. Billing periods shave the bill: quarters take 10% off, a year takes 20%. Every plan is unmetered on a 1-10 Gbps port and includes 1x IPv4 plus a /64 IPv6 subnet, so nightly multi-hundred-gigabyte syncs do not trip a traffic cap. Checkout is no-KYC, with BTC, XMR and USDT on seven networks among the accepted assets.

Pick an offsite backup box when — and skip one when

Pick an offsite backup box when your primary server holds data you cannot regenerate: client databases, mail archives, configuration states, git history. The rule that decides it is physical separation — a backup on the same host, or even the same rack, dies with the primary. A second BitCloudVPS in a different location gives you that separation at VPS prices: a Starter box is $8.50/mo list with 2 vCPU, 4 GB ECC RAM and 60 GB NVMe, and every plan carries the same unmetered 1-10 Gbps port, so a 40 GB restore does not turn into a bandwidth bill. It is also the right shape when you want the backup target itself unlinked from the primary account: token-only signup, no email required, and a payment in XMR keeps the two invoices from sharing an obvious trail.

Skip the second box when your real problem is capacity, not separation. If you need 1-5 TB of NVMe, the Storage line at $24-$56/mo is built for that, and a VPS plan topping out far below a terabyte will only disappoint you. Skip it too when the data is trivially reproducible — a stateless web frontend or a cache — because a nightly snapshot to object storage you already pay for beats a whole server doing nothing. And if your threat model is a compromised primary that reaches the backup over its stored credentials, know that no location choice fixes that: the fix is pull-based jobs, restricted keys and append-only targets, which is setup work, not hardware.

Backup math on the plan sheet: Starter, Plus and Business side by side

The direct answer: a single-server offsite target almost always lands on Starter or Plus, and the table below is the whole decision. All VPS plans ship NVMe with ECC RAM, one IPv4 and a /64 IPv6 block, so the columns differ only in compute and disk:

Two sizing habits make the tier obvious. First, keep the retention window under half the disk: on Starter's 60 GB that means roughly 30 GB of generations, which with deduplicating tools covers weeks of daily snapshots of a small server. Second, leave headroom for the restore working set — the fastest way to kill a backup box is a full-recovery job with no room to unpack. When one target serves two or three primaries, add their retained volumes together and step up a tier rather than compressing harder; at $13.50 for Plus the jump from Starter costs $5/mo list and doubles both RAM and disk. Quarterly billing takes 10% off each of those numbers and annual takes 20%, with Nano's $36/yr already priced flat.

Paying for backups without attaching your name to them

The checkout takes 19 assets: BTC, XMR, ETH, BNB, SOL, LTC, TRX, XRP, DOGE, DASH, ADA, GRAM on TON, and USDT on seven networks (TRC-20, ERC-20, BEP-20, Polygon, Arbitrum, Optimism, SPL). For a backup box the practical split is simple: XMR when the payment graph itself should stay opaque, since the primary and the target being paid from the same wallet is exactly the correlation you may want to avoid; USDT-TRC-20 when you want a stable USD price and low fees on a recurring invoice. Signup runs on a token with email optional, so the target account can exist with no mailbox tying it to the primary.

Terms work the same as everywhere on the network: monthly at list, quarter at −10%, year at −20%. On a Starter target that turns $8.50 into roughly $91.80 a quarter or $81.60 a year in list terms. If you would rather fund once and renew from balance, the tiers apply at $1,000 (+10%) and $2,000 (+40%) — worth it only if the balance will actually be consumed by renewals, because the credit is account-bound. One invoice note for automated setups: renewals debit the balance you preloaded, so a long retention plan is also a plan for keeping that balance topped up without an urgent on-chain payment at 3 a.m.

What an offsite backup box won't do for you

It will not verify your backups. A target holds whatever the sender pushes; if the job has been silently failing for six weeks, the box will faithfully store six-week-old data and no panel will tell you. Test restores on a schedule — that is client-side discipline, not a hosting feature.

It will not make the backup anonymous by location. A second server in Iceland paid from the same exchange account as a primary in Amsterdam is one invoice trail with two IPs; if unlinking matters, the payment wallet and the signup token have to differ too. It also will not protect the target from the credentials of the sender: a compromised primary with a stored key can delete or encrypt whatever the key allows, so restrict the key to append and prune on a window, or push from a read-only staging copy. Encryption is likewise your layer — client-side, before upload, with keys that never touch the box. And it will not outrun physics on restore: a 200 GB recovery across an ocean takes what the route takes, so pick the target region for where restores will actually be consumed, not for where the primary lives.

Splitting the 3-2-1 copy set across racks without doubling the bill

The classic 3-2-1 rule asks for three copies, two media, one offsite — and the offsite copy is where jurisdiction becomes a design input rather than an afterthought. Because every BitCloudVPS location prices identically, moving the backup copy to a different legal jurisdiction from your primary costs nothing extra: a primary in Amsterdam and its offsite copy in Reykjavik or Zurich pay the same plan sheet, and the choice of second rack is a pure risk decision. Pick the second rack for what threatens the first — different power grid, different legal system, different natural-disaster profile — not for latency, because backup traffic tolerates 80–230 ms of RTT without complaint.

Sizing is where backups get overbought. A single 60 GB Starter server with a realistic compression and retention ratio usually fits its incremental chain on a Nano's disk, and the flat $36/yr Nano price makes it the cheapest honest offsite target on the sheet. Add a second or third source server, or keep full rather than incremental chains, and you move up to Starter and its 60 GB NVMe, or Plus and Business above it. Restores ride the same unmetered 1–10 Gbps ports as everything else, so a full recovery is bounded by your own pipeline, not by an egress meter.

Two hygiene rules finish the design. Encrypt on the client before anything crosses the wire — the backup box should hold ciphertext only. And rehearse a restore quarterly, because an unrehearsed backup is a hope, not a copy.

Where the limits are (October 2026)

Keep the threat model precise. Anonymous-adjacent hosting removes the host-side identity trail: token-only signup, optional email, no KYC review, crypto invoices that record a txid rather than a name. It does not blind the host to traffic it must carry, it does not make a logged-in panel session anonymous, and it does not protect workloads that identify themselves — an admin login from your home IP defeats the whole design. Exact limits of no-KYC anonymity. OPSEC guide. Full checklist. Payment-side note: chain analysis happens on public blockchains, not at the host. Coin choice and network choice are your OPSEC knobs, and the checkout keeps 19 of them open.