Skip to content

Reliable IT system support and administration for businesses in Germany.

IT-Service Rottenkolber
EN
Disaster Recovery

Data Backup, Recovery and Disaster Recovery Planning

A backup is a copy of data, while recovery is the ability to bring systems back into working order. Treating these as one task often leaves organisations with copies that cannot be restored in time. This article explains how backup strategy and recovery planning work together to limit data loss.

By Editorial team of IT-Service Rottenkolber Published 8 min read

Key points

  • Test restores A backup is only trustworthy once a restore has been performed successfully and the results have been documented.

  • Keep the plan current A disaster recovery plan should name critical systems, priorities and activation criteria, and it must be reviewed after significant changes.

  • Assign clear roles Named roles with deputies, practised in regular exercises, prevent confusion and delay during an incident.

Separating Backup From Recovery Goals

Many organisations treat backup as a single technical job, yet the real question is what the business must be able to restore, and how quickly. Two measures help frame this: the recovery time objective, which describes how long a system may be unavailable, and the recovery point objective, which describes how much recent data may be lost. These targets are business decisions as much as technical ones, because they determine how much a day of downtime or a lost afternoon of work actually costs. Setting them first makes it possible to judge whether a given backup setup is adequate.

Once the targets are agreed, the backup design follows from them. A system with a tight recovery point objective needs copies taken more often than one that can tolerate losing a day of changes. Restore speed also matters, since a complete archive that takes days to unpack may not meet a short recovery time objective. Keeping backup and recovery goals apart in discussions tends to produce copies that exist but cannot be used when they are needed most.

Recovery goals should be reviewed whenever systems change, new applications are added or working patterns shift. A reporting database that was once updated weekly may now receive constant transactions, which silently breaks an earlier assumption about acceptable loss. Documenting the agreed targets next to each system makes such drift visible. A short review each year, or after any significant change, keeps the backup strategy aligned with what the organisation truly depends on.

Choosing Backup Frequency and Scope

The frequency of backups should follow from the recovery point objective rather than from habit or convenience. Full copies are thorough but heavy, so many setups combine periodic full backups with incremental or differential copies in between. The right balance depends on data volume, network capacity and how much storage the organisation is willing to retain. Scheduling backups outside peak working hours also reduces the chance that they slow down everyday operations.

Scope is equally important, because not every system needs the same treatment. Operating system images, application configurations, databases, user files and shared documents each change at different rates and carry different risks. Some data may be regulated or commercially sensitive and therefore require longer retention or additional safeguards. Listing what is included, what is deliberately excluded and why, gives a clear basis for audits and for later decisions.

Retention periods complete the picture. Keeping copies for a short window protects against recent mistakes but not against errors discovered weeks later, such as silent data corruption or a slowly spreading infection. A layered retention scheme, with daily, weekly and monthly copies, offers protection across different timescales. The schedule should be written down so that everyone responsible knows what is kept, for how long and where it resides.

  • Define recovery point and recovery time objectives for each critical system.
  • Combine full and incremental backups to meet the agreed frequency efficiently.
  • Schedule backup windows outside peak working hours.
  • Document what is included in the scope and what is excluded, with reasons.
  • Set layered retention periods covering daily, weekly and longer timescales.

Protecting Backups From Compromise

A backup that an attacker can encrypt, delete or alter provides little protection. Ransomware and insider mistakes frequently reach backup storage if it is permanently connected to the same network and credentials as production systems. Isolating copies, for example by keeping at least one version offline or in a separate account with restricted access, makes them far harder to compromise. Access to backup consoles should use strong authentication and be limited to a small number of named roles.

Immutability is a useful additional safeguard. Storage that prevents modification or deletion of backups for a defined period means that even someone with administrative rights cannot quietly remove the last good copy. Encryption of backups at rest and in transit protects the data if media are lost or intercepted. Keeping encryption keys separate from the backup files themselves avoids the situation where the protection and the protected data fail together.

Monitoring completes the protection. Alerts for failed jobs, unusual deletion activity or sudden changes in backup size can reveal a problem before it becomes a disaster. Regular review of access logs for the backup environment helps identify misuse early. Treating backup infrastructure as a critical system in its own right, with its own patching and security reviews, reflects the fact that it is often the last line of defence.

Testing Restores on a Regular Basis

Testing restores on a regular basis is the step that most organisations skip, yet it is the only way to know whether a backup can actually bring systems back. A backup job that reports success is not proof of recoverability, because files can be corrupted, encryption keys can be missing, or dependencies can be absent from the copy. A scheduled restore test, performed on an isolated system, reveals these gaps before a real emergency does. We recommend recording the time each restore takes, since that figure feeds directly into the recovery objectives defined elsewhere in your plan. Treat every test as a small incident drill that checks people as well as technology.

Restore tests should cover a representative mix of data, including databases, file shares, configuration files and at least one full system image. Rotate the scope over time so that every critical system is tested within a defined cycle, and document which items were verified and which were not. Compare the results against the expectations written in your recovery plan, and note any mismatch as a finding with an owner and a due date. Keep the test environment separated from production so that a faulty restore cannot overwrite live data or reconnect to the network carelessly. Over time, these records become the most reliable evidence that your backups work.

Many restore failures trace back to undocumented steps rather than broken media. When a test is performed by someone who did not configure the backup, unwritten assumptions quickly surface, such as credentials stored only on a single workstation. Each test should therefore be followed by an update to the runbook, so that the next person can complete the procedure without guesswork. Reviewing these updates periodically keeps the documentation aligned with changes in infrastructure. A tested and current runbook is worth far more during a crisis than a theoretically perfect strategy.

  • Restore a sample of databases and confirm application-level consistency.
  • Restore file shares and verify permissions, not only file contents.
  • Measure how long each restore takes and compare it with the recovery time objective.
  • Record every deviation as a finding with a responsible owner.
  • Update the runbook immediately after each test.

Writing a Disaster Recovery Plan

A disaster recovery plan turns the backup strategy into a structured response for serious disruption, such as a hardware failure, a fire, a ransomware infection or the loss of a data centre. It describes which systems are critical, in what order they must return, and which alternative resources are available if the primary site cannot be used. The plan should start from the business processes that depend on IT rather than from the technology alone, because this shows which outages are acceptable and which are not. Clear priorities prevent teams from spending scarce time on low-value systems while essential services remain offline.

A practical plan usually contains an inventory of systems with their owners, dependencies and recovery objectives, along with contact details for vendors and internal decision-makers. It sets out the activation criteria, meaning the conditions under which the plan is triggered and who is authorised to trigger it. It also describes communication steps for staff, customers and authorities where relevant, and it links to the restore procedures documented in your runbooks. Keep the plan concise enough to be read under pressure, and store printed copies offline, since an electronic version may be inaccessible during the incident.

A plan is only useful if it reflects the current environment, so it should be reviewed after significant changes such as new applications, migrations or organisational restructuring. Assign an owner who is responsible for keeping the document accurate and for scheduling reviews. Lessons from restore tests and from real incidents should flow back into the plan as concrete amendments rather than being filed away. Reviewing the plan together with the people who would execute it helps identify unrealistic assumptions early.

Assigning Roles During an Incident

During an incident, unclear responsibilities cost as much time as technical faults, because several people may act on the same system or nobody may act at all. A well-designed plan therefore names roles rather than individuals alone, so that responsibilities remain clear when staff are absent or change. A coordinating role should make decisions about activation, priorities and communication, while technical roles carry out the restore and verification steps. Separating these functions reduces the risk of conflicting actions and keeps the response traceable.

Typical roles include an incident lead who maintains an overview, a technical lead for each critical system, a communications contact for internal and external messages, and a documentation keeper who records actions and timestamps. Each role should have a named deputy so that the response can continue if the primary holder is unavailable. Access rights needed for these roles must be prepared in advance, because obtaining credentials during an emergency is a common source of delay. Write the role descriptions into the plan and confirm that the people involved understand them.

Roles only work when they are practised, so include them in restore tests and tabletop exercises at regular intervals. Exercises reveal overlaps, gaps and bottlenecks that no document can fully anticipate, and they build the confidence needed to act decisively. After each exercise, review the sequence of decisions and adjust the role assignments where necessary. Clear ownership, rehearsed regularly, is what turns a written plan into an effective recovery capability.

This article is general information and does not replace professional advice.

Related articles