BLOG / RECOVERY VALIDATION · TEKLink224
Backup testing: 7 steps to restore NAS, servers and SQL
When a drive fails, a folder is deleted or a server stops, a business needs usable data. A completed backup notification answers only part of that requirement.
Start the 7-step exercise ↓
From TRecovery by TUNGTEK’s data recovery perspective, a restoration exercise should establish what was recovered, its data point, whether the application works and the total time required. This guide outlines a repeatable test for Synology and QNAP NAS, servers, virtual machines and SQL Server.
Use this process proactively on stable systems. If a device already has read errors, a RAID member has failed or encryption is in progress, assess the condition and preserve the source before attempting restoration.
Validate a backup at three levels
A backup folder and a completed recovery are different milestones. The following acceptance criteria help establish what a test has actually demonstrated.
| Level | What to establish | Evidence to retain |
|---|---|---|
| Backup copy | The required version is readable, complete and accessible without the original system. | Version identifier, data point, inventory and read/restore logs. |
| Application | Files open, the database restores or the service runs with its dependencies. | File checks, database validation and application logs. |
| Business work | The responsible user validates the records, transactions or reports needed for operations. | Checks performed, reviewer and outstanding gaps. |
A hash matching a trusted reference can confirm that a transferred file is unchanged. It does not prove that the original content was correct, free from malware or sufficient to operate an application. A sample-folder restoration establishes results only for the tested folder and version.
7 steps for a useful restoration exercise
Choose the data needed to resume work
Start with a critical dataset such as customer records, designs or an accounting database. Define the required data point, acceptable time-based data loss (RPO) and target time to resume the service (RTO). Ask its users to specify acceptance checks.
Inventory the complete backup and restore tools
Record location, creation time, version and dependent backups. Check the restore software, accounts, catalogs and separately protected passwords or encryption keys. Confirm access if the source NAS or server is unavailable. One visible file does not establish a complete recovery set.
Prepare a separate recovery destination
Use healthy storage with sufficient room for restored data and application working space. Isolate testing from production and avoid conflicting addresses or machine names. Control network connections, scheduled jobs, email and synchronization in restored VMs to avoid unintended transactions.
Restore with the appropriate tool
Select the recovery point and restore into a new location. Retain the input backup and current data until validation finishes. Record errors and skipped objects. If the goal is acceptance of a whole service, expand sample testing into a service-level exercise.
Compare data and perform business checks
Compare inventories, sizes, timestamps and trusted reference hashes where available. Open important file types; inspect subfolders, Unicode names and permissions. In a test application, retrieve a record, open a document or generate a report. Record technical failures and business discrepancies separately.
Measure the time until the service is usable
Measure preparation, data retrieval, restoration, startup and validation as part of the recovery process. Record individual stages to locate delays. Compare the recovered data point with the RPO and elapsed service recovery time with the RTO. Download speed alone cannot represent the entire process.
Record results, assign fixes and repeat
Retain the tested version, scope, logs, reviewer, measured times and outstanding checks. Assign an owner and retest date to each failure. Set a schedule according to business criticality, and repeat after major changes to applications, capacity, encryption keys or backup destinations.
The Canadian Centre for Cyber Security recommends frequent backup/restore testing and addressing discovered issues. The specific schedule should reflect the organization’s requirements. [1]
Additional checks for NAS, servers and SQL
| Workload | Useful test | Common gap |
|---|---|---|
| Synology / QNAP NAS | Restore the required folder version to a separate destination; open sample files and validate user permissions. | Shared folders, applications and iSCSI LUNs are different scopes. A folder test does not validate an entire LUN. |
| Windows / Linux servers | Restore the data and configuration needed by a service, then test access from a test client. | Permissions, service accounts, certificates, configuration and dependencies may be outside the data backup. |
| VMware / Hyper-V | Restore in an isolated network, then test startup and the services inside the VM. | Booting alone does not establish application usability. Record service, data and dependency checks. |
| Microsoft SQL Server | Restore to a test database or server and validate integrity and business data. | The recovery point may require a backup chain and certificates/keys. RESTORE VERIFYONLY does not replace an actual restore. |
| USB / Cloud / Object Storage | Retrieve an earlier version into independent storage and time the process. | Include access, decryption keys, retrieval delays, destination capacity and transfer costs. |
Synology documents different methods for restoring Hyper Backup data and LUNs; select the method appropriate to the existing set. For QNAP and other platforms, consult documentation for the specific product and version. [4]
Microsoft states that RESTORE VERIFYONLY checks completeness and readability, without verifying the structure of all contained data. Restoring a database elsewhere also requires suitable file paths and, for encrypted databases, the necessary certificates or keys. Perform validation on the prepared test destination. [2] [3]
VM verification tools such as Veeam SureBackup can support testing where the infrastructure, features and license allow it. Match automated results to the organization’s application acceptance criteria. [5]
A practical backup testing record
Use a short record for each exercise. These suggested TRecovery fields can be adapted to the actual environment.
| Field | What to record |
|---|---|
| System and scope | Service, folders/database/VM and selected test data. |
| Backup set | Location, version, data point and dependent components. |
| Test environment | Destination, capacity, tool versions and isolation from production. |
| Targets | RPO, RTO and the files, reports or application workflows to validate. |
| Measured result | Recovered data point, stage timings, read failures, missing files or incorrect permissions. |
| Business validation | Reviewer, completed checks and untested items. |
| Follow-up | Failures, owners, deadlines and retest date. |
What to retain when a backup fails to restore
Preserve the backup set and logs before making changes. Record software versions, error codes, times and attempted actions. Identify alternative recovery points or independent copies. Retain older versions while their recovery value is assessed.
If the source device has already failed, formatting, recreating volumes, rebuilding RAID or repairing in place can change recoverable data. Stop repeated read attempts on a drive with unusual sounds or recurring disconnects and discuss a suitable preservation method. Export recovered data to independent storage.
Before contacting TRecovery, prepare the device type, symptoms, priority data, backup condition and attempted actions. Recovery prospects depend on the actual sources; capacity and the backup software name cannot establish a success rate.
VIET NAM Backup Day 30.11: share a tested restoration
For the VIET NAM Backup Day 30.11 community activity, TUNGTEK invites IT practitioners and data owners to exchange NAS, server, VM and SQL backup experience. A useful contribution is a restoration you tested: its scope, acceptance criteria, problems and follow-up changes.
Use diagrams or sample data when sharing. Remove customer details, accounts, keys and internal access information. The date 30.11 follows the community activity name and does not establish an official observance.
Each verification step can reveal a gap before a real recovery is needed. Pick one important system and record the results of your first exercise.
Backup testing questions
How often should we test a backup restore?
Choose a schedule based on criticality, data change and resources. Repeat after major changes to applications, storage, backup tools or encryption keys, and record the latest test date and scope.
Does opening a few files validate an entire NAS backup?
It validates the tested samples. Broader acceptance requires coverage of important data, permissions, versions, LUNs or applications as applicable, and documentation of what remains untested.
Should test restoration overwrite live data?
Use a separate destination in a controlled test environment to preserve sources and avoid disruption. Actual production recovery requires a transition plan appropriate to the incident.
Does a failed backup mean all data is lost?
There is insufficient evidence for that conclusion. Assess the error, other versions, independent copies and source condition. Retain the backup and logs; recovery prospects depend on the individual case.
References
- Canadian Centre for Cyber Security — Ransomware playbook
- Microsoft Learn — RESTORE VERIFYONLY
- Microsoft Learn — Restore a database to a new location
- Synology — Restoring Hyper Backup data and LUNs
- Veeam — Recovery verification for VMware vSphere
Read the prevention and community companion on Ransomware.VN.



