Off-site backups and a cold standby: protecting a VPS project from a provider outage

The short answer: keep at least one backup and one spare server with a different company than your main host, and keep your DNS with a third party, so an outage at the provider cannot take your data and your way back online with it. For a small project this costs about $6 to $11 a month. The reminder came on October 20, 2026, when Google Cloud reported a power event in its <em>europe-west2-b</em> zone in London: the zone was unavailable, the rest of the region saw elevated errors, and Google's incident ran for about eight hours across 38 products, including Compute Engine, Persistent Disk and Cloud SQL. A backup or snapshot stored inside the same infrastructure was no help to anyone who needed to restore elsewhere during those hours. The setup below works the same way on any VPS provider.

Why backups at the same provider are not enough

The rule most teams start from is 3-2-1: three copies of your data, on two different kinds of storage, with one copy off-site. For a VPS project the important word is off-site, and it should mean off-provider. Provider snapshots and the automatic backups bundled into a plan are useful for undoing a bad update or a deleted file, and our guide to VPS backups and snapshots covers how to use them. But they sit in the same company's data centres, behind the same control panel and the same billing account. When that provider has an outage, a locked account or a billing dispute, those copies are out of reach exactly when you need them.

So the plan has three parts: a copy of the data at a second company, a server at a second company that can take over, and a way to send visitors to that server without asking the failed provider for anything. None of them requires a second full-time production setup. The standby stays cold, which means it exists or can be created quickly, but it does not serve traffic until you switch to it.

Off-site backups and a cold standby: what they cost

For the data copy, object storage from an independent company is the simplest target. Backblaze B2, for example, lists storage at $6.95 per TB per month, keeps the first 10 GB free and includes egress up to three times the stored volume, so 50 GB of nightly database dumps and uploaded files costs about $0.35 a month. Send an encrypted dump of every database and an archive of the files that are not in your code repository, keep a rolling set of daily copies plus a few weekly ones, and make sure the credentials for this bucket are not stored only on the production server.

For the cold standby, pick a small plan at a second provider in a region close to your users. At is*hosting the Lite Linux VPS is $5.94 a month with 1 GB RAM, 20 GB SSD and 2 TB of bandwidth, and the Start plan is $10.19 with 2 GB RAM and 30 GB SSD; both include automated weekly backups. If you prefer not to pay for an idle server at all, an hourly-billed cloud works too: Hostman lists a 2 vCPU, 2 GB, 60 GB server from $7 a month, billed at $0.00956 an hour, so a standby you create only for a restore drill costs a few cents. Add the storage and the total for a small project lands at roughly $6 to $11 a month. Compare that with what eight hours offline means for you: a store taking $1,000 a day loses about a third of a day's sales. You can find comparable small plans in other regions with the VPS calculator.

DNS outside both providers, and a restore you have tested

The switch itself happens in DNS, so the DNS zone should not live at the provider that may go down. A separate DNS host, such as the DNS included in Cloudflare's free plan, lets you point your domain at the standby while the main provider is unreachable. Set a short TTL in advance on the records you would change: Cloudflare allows DNS-only records down to 60 seconds on non-Enterprise plans, and its Auto setting is 300 seconds. A TTL of 5 minutes means most visitors follow the change within minutes instead of the hours a default TTL can take.

Finally, a backup you have never restored is a guess. Once a quarter, create or start the standby, restore the latest dump and files onto it, open the site through a test hostname, and write down how long it took and what was missing. The steps are the same as a planned move, which our guide on how to migrate to a VPS walks through, and the measured time is your real recovery time, not the one in a provider's marketing. Read your provider's SLA with this in mind: as our guide to VPS uptime and SLAs explains, an SLA usually pays service credits on your bill, while the cost of the downtime stays with you.