A reliable backup must preserve more than application files: it needs the service’s data, configuration, deployment details, and any credentials or encryption keys required to restore them. Keep copies outside the server’s failure domain, document how to rebuild the service, and test that process on a clean target. The exact backup and restore steps depend on your application, database, and hosting setup; no single file-copy command safely covers every self-hosted service.
What to back up for each service
Think in terms of a recovery set: everything needed to rebuild the service and return it to a usable state. Make an inventory for each service before choosing backup software or scheduling jobs.
As an Amazon Associate I earn from qualifying purchases.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →- Persistent application data: Identify the directories, volumes, or storage mounts that hold uploads, documents, media, and other user-created content.
- Database: Record the database engine, where it runs, and the application-specific procedure for creating a consistent backup and restoring it. Copying database files while the database is active may not produce a usable backup; follow the database and application documentation.
- Configuration and deployment: Preserve configuration files, compose or other deployment definitions, service settings, and notes about required software versions and dependencies.
- Secrets and recovery material: Record how to retrieve passwords, tokens, certificates, and encryption keys without putting sensitive values in an unprotected backup or an undocumented location.
- External dependencies: Note any data or services outside the host, such as a separately hosted database, network storage, or DNS settings. Decide how each is protected and restored.
Mark which copy of each item is authoritative and how it fits into the restore order. A platform-level backup can capture substantial machine state, but it does not automatically prove that every application-specific secret or external dependency is included. Proxmox VE, for example, describes its integrated guest backup as including guest configuration and data; treat that as a Proxmox-specific feature, not a guarantee for arbitrary services or other platforms (Proxmox VE backup and restore features).
Choose a backup approach that fits the service
Whole-machine or virtual-machine backups and service-aware backups solve different problems. Some setups use both: a machine-level copy for broad recovery and application- or database-aware backups for more precise recovery. Compare the options against your service’s consistency requirements, downtime tolerance, and ability to restore onto replacement hardware.
#1 Best Overall
- Easily store and access 2TB to content on the go with the Seagate Portable Drive, a USB external hard drive
- Designed to work with Windows or Mac computers, this external hard drive makes backup a snap just drag and drop
- To get set up, connect the portable hard drive to a computer for automatic recognition no software required
- This USB drive provides plug and play simplicity with the included 18 inch USB 3.0 cable
- The available storage capacity may vary.
| Approach | Useful for | Questions to check |
|---|---|---|
| Whole-machine or VM backup | Recovering a guest or system with its operating-system and guest state. | Does the backup include all required attached storage and configuration? Can it be restored to different hardware or a new host? Is the application’s data consistent at the time of backup? |
| Service-aware file and database backups | Restoring selected services, files, or database state without necessarily restoring an entire machine. | Does the procedure produce a consistent database backup? Are uploads, configuration, and secrets included? Are the database and application versions compatible with the restore? |
| Combined approach | Keeping a broad recovery route alongside more targeted restores. | Are the copies independent, and have both restore paths been tested? Does either rely on the other’s credentials, storage, or host? |
Compare actual restoration speed and granularity, data consistency, portability, off-site transfer needs, encryption and key recovery, resistance to compromised production credentials, and how easily you can test a restore. Proxmox documents features such as live restore and single-file recovery for supported Proxmox Backup Server VM backups; those features are specific to that platform and do not remove the need to meet each application’s consistency requirements (Proxmox VE features).
Keep copies in separate failure domains
A backup on the same disk or storage system as the live service can disappear with it. Proxmox Backup Server documentation describes the 3-2-1 rule as three copies of the data, on at least two different storage media, with one copy off-site (Proxmox Backup Server storage documentation). Use it as a resilience principle, not as a substitute for checking that the copies are independent and restorable.
Rank #2
- Easily store and access 5TB of content on the go with the Seagate portable drive, a USB external hard Drive
- Designed to work with Windows or Mac computers, this external hard drive makes backup a snap just drag and drop
- To get set up, connect the portable hard drive to a computer for automatic recognition software required
- This USB drive provides plug and play simplicity with the included 18 inch USB 3.0 cable
- The available storage capacity may vary.
- Keep at least one copy on separate infrastructure from the production server.
- Keep an off-site copy for events that affect the whole location, such as theft, fire, or a site-wide outage.
- Consider whether the account or credentials used by the live server can delete or overwrite every backup copy.
- If you use an external drive, disconnect or otherwise protect it when it is not being updated. A drive kept beside the server can provide a separate local copy, but it is not an off-site copy.
Set backup timing and retention from your recovery needs
Choose how often to back up by deciding how much recent change you could afford to lose. Choose retention by deciding how far back you may need to recover—for example, if a problem is discovered after it has already been copied into newer backups. These are service-specific decisions; there is no universal schedule or retention period established for every self-hosted setup.
Also decide how quickly each service must be back online. A service that can tolerate a slower rebuild may need a different recovery arrangement from one that supports essential work. Write down the acceptable data loss and downtime for each service, then check that the chosen backup frequency, storage, and restore procedure can meet those expectations.
Rank #3
- Easily store and access 1TB to content on the go with the Seagate Portable Drive, a USB external hard drive.Specific uses: Personal
- Designed to work with Windows or Mac computers, this external hard drive makes backup a snap just drag and drop. Reformatting may be required for Mac
- To get set up, connect the portable hard drive to a computer for automatic recognition no software required
- This USB drive provides plug and play simplicity with the included 18 inch USB 3.0 cable
- The available storage capacity may vary.
Protect encrypted backups and their keys
Encryption can protect sensitive backup contents, but it also makes key recovery part of disaster recovery. Proxmox Backup supports client-side AES-256-GCM encryption and can use a master key to keep an encrypted copy of the backup key with the backup (Proxmox Backup client usage and encryption; Proxmox Backup command syntax and key recovery). This describes Proxmox Backup, not every backup product.
Keep an independent, recoverable copy of the private key or other recovery material needed for your setup. Confirm who can access it and how it will be obtained if the original server and its credentials are unavailable. Test the recovery method; an encrypted archive that cannot be decrypted is not a usable backup.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteRank #4
- Easily store and access 4TB of content on the go with the Seagate Portable Drive, a USB external hard drive.Specific uses: Personal
- Designed to work with Windows or Mac computers, this external hard drive makes backup a snap just drag and drop
- To get set up, connect the portable hard drive to a computer for automatic recognition no software required
- This USB drive provides plug and play simplicity with the included 18 inch USB 3.0 cable
- The available storage capacity may vary.
Document the clean-host restoration order
Write a restore runbook while the service is operating, when you can still inspect its dependencies. Keep the runbook somewhere accessible if the server fails, and avoid making the only copy dependent on the host or account you are trying to recover.
- Prepare a replacement target. Record the operating system or hypervisor requirements, storage mounts, network needs, and software dependencies. Use an isolated or separate target for testing when possible.
- Retrieve the backup and recovery material. Locate the correct backup, credentials, encryption keys, configuration, and deployment definition. Confirm you can access the backup storage from the replacement environment.
- Recreate the service environment. Install or configure the required host, runtime, storage, and supporting services in the order documented for your stack.
- Restore the database and application data. Follow the application’s and database’s supported restore procedures. Restore related pieces to compatible points in time where the software requires it.
- Apply configuration and secrets. Restore the settings and credentials the service needs, taking care not to expose them in logs or an unprotected test environment.
- Start dependent services in order. Bring up databases and other prerequisites before the application when the stack requires that sequence.
- Verify the result before returning traffic. Check that the service starts, login works, representative records and files are present, and logs do not show unresolved errors. Confirm any external dependencies are reachable.
The exact commands and sequence vary by service, database, runtime, and storage layout. Consult the official documentation for the specific application and database rather than assuming a generic file restore is safe.
Test restores, not just backup jobs
A successful backup job shows that a process ran; it does not establish that the data can be decrypted, restored, and used. Periodically restore a backup to a clean target and check both the service and its data. Proxmox Backup Server documentation recommends restore testing and notes that automation or sampled restores can help where testing every guest is impractical (Proxmox Backup Server restore testing).
Recommended Free Tools
If a full restore of every service is too costly, prioritize the most important services and rotate which ones you test. Record what was restored, what failed, how long the work took, and what must change in the runbook. The documentation does not set a universal testing interval, so choose one that fits how often your services and recovery setup change.
Quick Recap
Common backup and recovery failures to guard against
- Only the data directory was saved: The application may still need configuration, deployment details, database state, or credentials to start.
- Every copy depends on one host or account: A server failure or compromised credential may make all copies unavailable or deletable.
- The backup is encrypted but the key is missing: Store and test recovery material independently of the machine being protected.
- The database copy is inconsistent: Use the supported database- or application-aware procedure instead of assuming a live file copy is valid.
- The archive exists but no replacement environment is documented: Record dependencies and restore order, and prove the process on a clean target.
- The restore was never checked in the application: Confirm logins, representative content, and logs rather than stopping when files have been copied.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
