DigitalOcean backups vs snapshots: what should you pay for?
Choose DigitalOcean backups, snapshots or both with dated cost examples, database and volume limits, and a restore checklist for a self-hosted server.
Advertising disclosure: We earn commissions through the DigitalOcean affiliate button in this guide. Documentation links are ordinary source links.
For a self-hosted server that changes every day, start by pricing automated backups. Use a snapshot for a deliberate checkpoint before an upgrade. A snapshot you remember to take once is a poor substitute for a recovery schedule.
Neither purchase finishes the recovery job. Database exports, attached storage and a successful restore matter too. This guide helps you choose the paid disk-image layer for an existing or planned Droplet. It is based on vendor documentation, not a hands-on recovery test.
Choose by the failure you need to recover from
| Your job | Starting choice | What you still need |
|---|---|---|
| Recover a changing server after a bad update or deletion | Automated backups | A suitable interval, retention and a tested restore |
| Preserve a known state before a planned upgrade | An on-demand snapshot | A consistent application state and a deletion date |
| Recover a database with frequent writes | Application-level database backup first | Disk images if useful, plus a database-specific restore test |
| Protect files on an attached volume | A separate volume recovery plan | Volume snapshots or an appropriate file backup |
| Rebuild a disposable server entirely from code | Possibly neither disk-image product | Proof that code, secrets and persistent data exist elsewhere |
If losing yesterday's edits would be unacceptable, weekly backups are too infrequent for your requirement. Write down how much lost work you can tolerate before choosing a plan. Also decide how long recovery may take; buying a more frequent image does not prove a faster restore.
For a Home Assistant server, separate its application backup from the server image. For a Docker Compose app, inventory the database, uploads, configuration and secrets first. The disk-image choice comes after that inventory.
What you are buying
DigitalOcean's backup feature documentation describes scheduled, crash-consistent images taken while the Droplet runs. They capture disk state, not a promise that an application finished its current transaction. Backups stay in the Droplet's datacenter and do not include attached volumes.
Snapshots are on-demand images. They suit a named checkpoint that you want to manage yourself. Taking them through automation is possible, but then scheduling, monitoring and cleanup become your jobs.
For a planned snapshot, DigitalOcean recommends powering down the Droplet so applications can write their data to disk. Schedule the outage and stop services cleanly. A live snapshot can affect performance and data consistency.
Budget the recovery layer separately
Prices checked October 4, 2026, in US dollars before tax, currency conversion and other services. These are arithmetic examples, not a quote for a particular Droplet.
The basic backup price schedule adds 20% of the Droplet's monthly cost for weekly backups or 30% for daily backups. On an assumed $12 monthly compute bill, that means:
| Basic plan | Backup charge | Compute plus backup |
|---|---|---|
| Weekly | $12 × 20% = $2.40 | $14.40 per month |
| Daily | $12 × 30% = $3.60 | $15.60 per month |
Usage-based backups have a different calculation based on restorable file size, frequency and retention. Do not apply the percentage formula to that product. Enter the intended settings in the provider's calculator and record the resulting estimate. More retained recovery points change the comparison.
Droplet snapshots cost $0.06 per GB per month. One snapshot with a billable size of 20 GB would cost $1.20 for a full month; three such snapshots would cost $3.60. These examples assume equal sizes and full-month retention. Volume snapshots use GiB rather than GB, and the documentation lists a $0.01 minimum charge.
The $1.20 snapshot example buys one manually managed point, not the same service as scheduled daily backups. Add application backup storage, volume protection and temporary recovery-server charges to your budget. Count the time you spend maintaining a snapshot script too.
Check the limits before enabling backups
DigitalOcean's backup limits say backups are not encrypted at rest, though they are not externally accessible. The provider also cautions about high-write workloads and data still in memory or cache. For a busy database, use its supported backup mechanism rather than assuming a running-server image is sufficient.
Do not buy automated Droplet backups as your only protection if the important data sits on a volume. Do not treat an image in the same datacenter as an independent offsite copy. If those limits conflict with your requirements, design a separate recovery method before paying for this layer.
If a scheduled disk image fits your workload, check the selected Droplet's backup options and complete estimated bill in DigitalOcean. Existing customers can use their current control panel; opening another account is unnecessary. The button below is an affiliate link for readers who decide DigitalOcean fits their hosting needs.
Prove recovery before you rely on it
Use DigitalOcean's create-or-restore instructions to rehearse recovery into a separate Droplet. Keep the original server intact during this check. Budget the extra compute and isolate the copy from live traffic, outbound notifications and production integrations.
- Record the image date and the application backup that belongs with it.
- Bring up the recovery copy and restore any separately stored database or files.
- Check a recent record, an uploaded file and an authenticated application task.
- Confirm the copy does not run duplicate scheduled jobs or send real customer messages.
- Record elapsed recovery time and missing steps. Keep the recovery notes outside the server.
- Remove temporary resources when the rehearsal is complete, after confirming what must be retained.
A failed rehearsal is useful evidence. Fix the missing data or process before adding another paid recovery product. For an application-level example, the n8n restore guide covers the work a disk image alone does not settle.
My default for a small changing server is scheduled backups plus a deliberate pre-upgrade snapshot, with application backups where needed. For a disposable server, I would first test rebuilding it. Pay for the recovery path you can actually use.