Backup planning on a small VPS usually fails in one of two directions: no tested restore at all, or a storage bill larger than the server it protects. The right budget sits between those, set by two numbers most small sites never calculate — how much data changes per day, and how quickly you need it back. This article sizes both.
Start with RPO and RTO, not with a tool
The recovery point objective is how much data you are willing to lose, expressed in time. A brochure site that changes twice a month has an RPO measured in days. A site taking orders has an RPO measured in hours, because every lost hour is lost transactions. The recovery time objective is how long you can be offline while restoring, and it drives far more of the cost than the RPO does.
Writing these two numbers down first prevents the most common over-spend: paying for hour-by-hour retention on a site that only needs daily, because the tool’s default was hourly. It also prevents the more dangerous under-spend, which is daily backups with a four-hour manual restore on a site that earns money per hour.
Measure the change rate before choosing a retention window
Retention cost is driven by what changes, not by what exists. A 20 GB site where only 300 MB changes per day does not need 20 GB per day of storage. Measure actual change volume with a week of rsync --dry-run --stats output, or with file-count and byte deltas from an incremental snapshot.
Typical figures for small business sites: 0.5–2% daily change for a content site with a media library, 3–8% for a WooCommerce store with orders and logs, and much higher during plugin updates or migrations. Budget the worst realistic day, not the median, because the moment you need a backup is usually the moment a large change happened.
The four storage tiers and what each costs you
| Target | Typical cost | Restore speed | Right for |
|---|---|---|---|
| Snapshot on same host | $0–2/mo | Minutes | Rollback after a bad deploy |
| Second local disk / attached volume | $2–5/mo | Minutes to an hour | Fast recovery of config and databases |
| Object storage, same provider | $1–3 per 100 GB/mo + egress | 30 min to a few hours | Offsite copies of full archives |
| Object storage, different provider | $2–6 per 100 GB/mo + egress | Hours | Surviving a provider-level outage |
Snapshots are not backups. They share the failure domain of the host and are usually deleted with the account. The budget should treat snapshot cost as convenience spend and offsite object storage as the actual insurance line — the same way a shop keeps a fireproof safe somewhere other than the shop.
For most small sites, cost stacks as roughly $2 for snapshots plus $3–8 for offsite storage, against a $9 server. That is a 40–90% overhead on the hosting line, which is worth stating plainly in the budget rather than discovering at the end of the year.
Retention policy: grandfather-father-son in practical terms
A GFS rotation keeps 7 daily, 4 weekly and 6 monthly copies — 17 restore points for a storage cost of roughly a quarter of holding 17 full copies, because incrementals only carry the delta. On a 20 GB site with 2% daily change, that is about 26 GB total rather than 340 GB.
The trade is restore complexity: a restore needs the last full plus each increment in the chain, so a broken chain costs more than the storage saved. Choose tooling that verifies chain integrity automatically and alerts on failure. A backup that silently stopped three weeks ago is worse than no backup, because you have budgeted for it and still have nothing.
Egress is where the hidden bill lives
Object storage bills for retrieval, and a full restore of a 20 GB site pulls 20 GB at once. At typical egress rates of $0.01–$0.09 per GB, that is $0.20 to $1.80 per restore — trivial. The expensive case is a restore loop during a long outage, or storing backups on the same provider you host with and paying egress when you leave. Check both rates before committing.
Budget an annual full-restore test explicitly, not just storage. One test costs a few dollars and one afternoon; discovering a broken chain during an outage costs the outage itself. Our budget VPS plans list included storage and bandwidth so the egress side of this arithmetic is visible before you start.
A worked monthly budget for three site types
| Site type | Data | Daily change | Plan | Storage cost |
|---|---|---|---|---|
| Brochure, 10 pages | 3 GB | 20 MB | Daily, 14 days + monthly | $1 |
| Content site with media | 25 GB | 400 MB | GFS: 7/4/6 | $4 |
| Small store, 300 SKUs | 40 GB | 1.8 GB | Hourly DB + daily GFS | $9 |
The store pays for hourly database dumps because order data is the only truly irreplaceable asset; media and plugin files can be re-downloaded or re-uploaded. Splitting the backup target by data class — database hourly, files daily — usually cuts total storage 30–50% versus backing everything up on the same schedule.
Apply that split before increasing the retention window. Most small sites asking for “more backups” actually need a correct schedule split and a restore that has been tested once.
What to put in the budget template
Four lines: storage at target retention, snapshot convenience tier, expected annual restore tests, and a contingency for growth of the media library. Review the change-rate number yearly, because a site that grows a video library can move from 2% to 15% daily change without anyone noticing.
For a small business site the realistic total is $3–$12 a month on top of hosting — cheap relative to the cost of a multi-day outage, and cheap relative to the embarrassment of telling customers the last good copy was eight weeks old. Check current plan storage allowances on our cheap VPS plan comparison and size the offsite line from there.

