A backup that has never been restored is a hope, not a control. Disaster recovery testing replaces that hope with evidence: a record of what was restored, how long it took, what data was missing and who accepted the result. The checklist below is a working document, not a maturity questionnaire. Copy it into your operations manual, assign the owners named in each section, and keep the completed record with the rest of your audit evidence.
It assumes a handful of critical services rather than a dedicated continuity team, which fits a retailer, a distributor and most service firms. Adapt the service names, keep the structure, and do not skip the sign-off.
How to use the checklist
Run three levels of test. Monthly, confirm that backup jobs completed and a sample file restores. Quarterly, restore one critical service into an isolated environment and measure it. Annually, exercise the full sequence with the people who would actually be called at night. Each section names an owner, frequency, evidence and sign-off.
| Test level | What it proves | Frequency | Record |
|---|---|---|---|
| Restore verification | The backup set is readable | Monthly | Job log and checksum |
| Component restore | One service can be rebuilt | Quarterly | Test log and reconciliation |
| Full exercise | People, dependencies and communications work | Annually | Signed exercise report |
1. Scope and service inventory
Owner: IT operations. Frequency: reviewed quarterly and after any platform change. Evidence: a signed service register with a named owner for each row. Sign-off: operations lead.
List every service the business needs in order to trade: storefront, order management, payment capture, email, DNS, file storage, accounting, HR, the deployment pipeline and SaaS tools holding customer records. For each, record the backup method, repository, retention, encryption key owner and restore procedure, and mark services holding personal or payment data. Without this register, the test covers the easiest system rather than the one that matters.
2. Recovery objectives and decision rights
Owner: the business owner for each service. Frequency: annually and whenever a revenue system changes. Evidence: an objective table with approval dates. Sign-off: business owner, not IT alone.
For each service, write a recovery point objective and a recovery time objective in plain language. Add the acceptable degraded mode, such as taking orders by phone while the site is down. Name the person who declares a disaster, who authorises emergency spend and who speaks to customers. These are commercial decisions, so they cannot be delegated entirely to the technical team.
3. Pre-test readiness
Owner: backup administrator. Frequency: before every component restore and full exercise. Evidence: a completed readiness ticket. Sign-off: test lead.
Confirm that the last thirty days of backup jobs succeeded and the repository is reachable from the test network. Check that encryption keys and recovery passwords are available to the test team, not locked inside one person's password manager. Reserve isolated compute and storage, confirm that licences allow a restore, and freeze unrelated changes. Tell support and finance what will happen so a test alert is not treated as a real incident.
4. Restore record and application checks
Owner: test lead. Frequency: every test. Evidence: a timestamped test log with screenshots and command output. Sign-off: test lead and application owner.
Record the start and end time, the backup set identifier, the recovery point and the operator. Note each step and any deviation from the runbook. Measure recovery from the decision to restore, not the first command. Then test the application, not only the database: log in, open a recent order, issue a sandbox refund, send a test email and confirm the payment callback still works.
5. Data integrity and reconciliation
Owner: finance with the application owner. Frequency: every component restore. Evidence: a reconciliation sheet signed by both. Sign-off: finance lead.
A restored system can open while the numbers are wrong. Compare record counts and financial control totals between production and the restored copy. Check the latest order, invoice, payment and stock movement, and look for sequence gaps, missing attachments and broken Bangla text. Confirm that user accounts and permissions match the intended state, not a former employee's access. A test that cannot reconcile totals is a partial pass.
6. Security, privacy and deletion
Owner: security lead. Frequency: every test that uses live data. Evidence: an access list, credential rotation record and deletion confirmation. Sign-off: security lead.
The test copy contains live personal data. Limit access to named testers, mask customer records where possible, and use temporary credentials rotated afterwards. Log who accessed the copy. When the test ends, delete the environment and record the deletion; if a vendor holds the backup, confirm deletion in writing. This stops a resilience exercise from becoming a privacy incident.
7. Findings and remediation
Owner: operations lead. Frequency: reviewed monthly until closed. Evidence: a findings register with severity, owner, due date and retest date. Sign-off: operations lead and business owner.
Every gap becomes a finding: missing credentials, an undocumented dependency, a runbook that assumes one person, or a backup that excludes a table. Do not close a finding because a retry passed without a change. Track the age of open findings and escalate anything overdue. A test that fails and produces a funded fix is more valuable than a passing test that proved nothing.
8. Annual exercise and sign-off page
Owner: operations lead with all service owners. Frequency: annually and after a major platform change. Evidence: a one-page exercise report. Sign-off: managing director or delegated business owner.
Keep one page for every test: scope, date, participants, result, exceptions, next test date and signatures. Include the contact tree, out-of-band channel, vendor escalation numbers and customer notice template. Test the call tree and record how long it took to reach each person. Store the page with the findings register and updated runbook, and link it from the service register.
Metrics for the management pack
Track actual recovery time and data loss against target, the share of in-scope services tested in the last twelve months, open findings by age and restores that needed vendor help. Define each measure and show the trend, not a single result. These numbers show whether recovery is improving or whether the same small file is restored every quarter to produce a clean report.
Common ways a test proves nothing
Restoring the same small file every quarter, restoring onto the production host, accepting a successful backup job as proof of recovery, skipping finance sign-off and never updating the runbook are the usual shortcuts. Each leaves a record that looks complete. A reviewer should be able to open the evidence pack and see the backup identifier, the recovery point, the measured time and the reconciled totals without asking anyone.
Recovery testing sits within Managed Technology & Cybersecurity, alongside monitoring and incident response. Pricing explains how a test or retainer is scoped. For the budget side of the same decision, read what backup and disaster recovery testing costs in Bangladesh.
Sources
- National Institute of Standards and Technology, SP 800-34 Rev. 1, Contingency Planning Guide for Federal Information Systems, accessed 17 November 2026.
- National Institute of Standards and Technology, SP 800-61 Rev. 3, Incident Response Recommendations and Considerations for Cybersecurity Risk Management, accessed 17 November 2026.
- Amazon Web Services, Reliability Pillar, AWS Well-Architected Framework, accessed 17 November 2026.
- International Organization for Standardization, ISO/IEC 27001:2022, Information security management systems, accessed 17 November 2026.
Responses (0)
No comments yet. Add the first one below.
Leave a response
Your email address will not be published. Required fields are marked *.