Backup Terminology: RPO, RTO and Disaster Recovery Glossary

The 3-2-1 rule is a backup principle that recommends keeping at least 3 copies of data, on 2 different storage media, with at least 1 copy stored at a different location. Production data counts as one of the three copies. The goal is to stop a single failure or incident from destroying all copies at once.

What it means for a business: A backup kept on a NAS in the same server room is lost together with production data in a fire or theft. Cloud backup is the most practical way to meet the off-site copy requirement.

Related: 3-2-1 backup rule · 3-2-1-1-0 rule · cloud backup

The 3-2-1-1-0 rule extends the 3-2-1 rule for the ransomware era. It adds that at least 1 copy should be offline, air-gapped or immutable, and that restore verification should show 0 errors. The last element makes regular backup testing mandatory.

What it means for a business: The rule stresses that a backup must not only exist but also survive an attack and actually open. Immutable cloud backup and automated verification are common ways to meet these two extra requirements.

Related: 3-2-1 rule · immutable backup storage · restore testing

An air gap is the physical or logical isolation of a backup copy from the production network. A physical air gap uses tapes or disks removed from the network; a logical air gap relies on separate credentials, network separation and access restrictions. The goal is to keep attackers from reaching the backup over the network.

What it means for a business: One of the most common weaknesses is that an attacker who takes over production can reach the backups with the same privileges. An isolated copy is one way to meet the extra “1” in the 3-2-1-1-0 rule.

Related: 3-2-1-1-0 rule · ransomware · ransomware resilience check

Application-aware backup is a method that brings applications such as databases or mail servers into a consistent state before the backup is taken. On Windows this is usually done with VSS (Volume Shadow Copy Service), which flushes pending writes to disk. For some applications, transaction log truncation also happens at this step.

What it means for a business: A SQL Server or Exchange backup taken without application awareness may open, yet the database inside can be inconsistent. For ERP and accounting systems, this setting directly affects whether a restore succeeds.

Related: backup management · restore testing

Asynchronous replication is a type of replication in which a write is acknowledged once it completes at the source, and changes reach the target after a short delay or at set intervals. The application does not wait for the target. The target copy may lag the source by seconds to hours.

What it means for a business: Because it works across distant sites and over standard internet links, asynchronous replication is the most common method in disaster recovery. The lag between copies corresponds to the RPO that has been accepted.

Related: synchronous replication · disaster recovery · RPO

BaaS (Backup as a Service) is a backup model in which the backup software, the storage and often the management are delivered by a service provider. Backups travel over the network to storage in the provider’s data center. Pricing is usually based on the number of protected systems or the capacity used.

What it means for a business: BaaS moves backups off the company premises automatically and removes the need to buy backup hardware. Where the data is stored, the retention options and restore support are the deciding factors when choosing a provider.

Related: cloud backup · backup management · DRaaS

A backup is a copy of data and systems stored separately from the production environment. Its purpose is to return data to its state at a specific point in time after deletion, corruption, hardware failure, ransomware or a disaster. A backup is only as valuable as the ability to restore it.

What it means for a business: Backup works like an insurance policy: the cost shows up every month, while the benefit appears only when something goes wrong. The way it is designed directly decides how much data can be lost and how long work stops.

Related: cloud backup · backup strategies · 3-2-1 rule

Business continuity is an organization’s capability to keep its critical activities running at an acceptable level during a disruption or crisis. Beyond IT systems, it covers staff, facilities, suppliers and communication. ISO 22301 is the international standard for business continuity management systems.

What it means for a business: Disaster recovery is the IT part of business continuity; even if servers come back, work stays stopped without a plan for people, processes and communication. A business impact analysis decides which processes the plan prioritizes.

Related: business impact analysis · disaster recovery · ISO 22301

A business impact analysis (BIA) identifies the operational and financial effects that would follow if business processes, and the systems supporting them, were disrupted. NIST SP 800-34 describes the BIA as a key step in the contingency planning process. The analysis produces a criticality class and RTO and RPO targets for each system.

What it means for a business: Without a BIA, every system gets the same protection: either money is wasted or a critical system is underprotected. The hourly cost of downtime is the most tangible output of the analysis.

Related: downtime cost calculator · RTO · business continuity

Deduplication is a storage technique in which repeated blocks in backed-up data are stored only once, and other copies are represented by references to that block. It can run at the source, on the backup server or on the target storage. It is usually combined with compression.

What it means for a business: Across many virtual machines running the same operating system, deduplication can noticeably reduce the backup space and network traffic required. The ratio achieved depends on the data type; gains are low on encrypted or already compressed data.

Related: backup storage calculator · cloud backup

A differential backup copies all data that has changed since the last full backup. Because it ignores earlier differentials, each new differential is larger than the one before it. A restore needs only the last full backup and the latest differential.

What it means for a business: Differentials use more space than incrementals but keep the restore chain short. Where full backups are infrequent, storage planning has to account for differentials growing quickly over time.

Related: full backup · incremental backup · backup storage calculator

Disaster recovery (DR) is the process of bringing IT systems and data back into operation at a secondary location after a data center failure, fire, flood, cyberattack or similar major event. It combines a plan, technology and regular testing. Its targets are defined by RPO and RTO.

What it means for a business: Unlike restoring a single file, disaster recovery means running an entire service somewhere else. Without a written plan and a tested switchover procedure, technology alone is not enough.

Related: disaster recovery service · how to create a disaster recovery plan · business continuity

DRaaS (Disaster Recovery as a Service) is a disaster recovery model in which systems are replicated to a service provider’s cloud infrastructure and run there during a disaster. The secondary data center, hardware and replication platform are operated by the provider. It is usually billed on a subscription basis.

What it means for a business: DRaaS removes the capital cost and operational burden of building a second data center. The contract should state RPO and RTO targets, testing frequency and failover responsibilities clearly.

Related: disaster recovery (DRaaS) · failover · BaaS

Failback is the process of moving workloads back to the original environment once the primary system or site is ready again after a failover. It includes synchronizing changes made while running at the secondary site back to the primary. It is preferably done in a planned maintenance window.

What it means for a business: Failback often takes longer than failover, and if data synchronization is skipped, records entered during the disaster can be lost. A disaster recovery plan should cover the return steps as well.

Related: failover · data replication · disaster recovery plan

Failover is the transfer of workloads to a standby system or secondary site when the primary system or location becomes unavailable. It can start automatically or manually with administrator approval. Network routing, IP addresses and DNS records must also be updated for the new location.

What it means for a business: Failover is the moment that shows whether a disaster recovery plan exists beyond paper. Who decides, which systems start in which order and how users are redirected should all be written down in advance.

Related: failback · disaster recovery · restore testing

Forever forward incremental is a method in which one full backup is created at the start of the chain and only incrementals follow. When the retention limit is reached, the oldest incremental is merged into the full backup, keeping the chain at a fixed length. Veeam Backup & Replication documents it among its backup methods.

What it means for a business: Production systems are not read in full again and again, which reduces load on servers and the network. The merge step, however, creates read and write load on the backup repository, so storage performance has to be chosen accordingly.

Related: synthetic full · retention · backup management

A full backup copies all data in the selected source completely, independent of any previous backups. Because a restore needs only one file set, it offers the simplest recovery path. In return, it takes the longest time to create and uses the most storage space of all backup types.

What it means for a business: Taking a full backup every day on large data sets stretches both the backup window and the storage bill. Full backups are therefore usually taken weekly or monthly, with incremental backups covering the days in between.

Related: incremental backup · synthetic full · backup storage calculator

GFS (Grandfather-Father-Son) retention is a tiered scheme in which restore points are kept for different lengths of time. Daily backups are kept briefly, weekly backups longer, and monthly and yearly backups the longest. This covers a long time span with a small number of backups.

What it means for a business: GFS makes it possible to bring back both yesterday’s file and a record from months ago without letting storage costs explode. Long retention needs driven by audits and regulation are usually met with the monthly and yearly tiers.

Related: retention · KVKK and data backup · backup storage calculator

An immutable backup cannot be deleted, overwritten or encrypted for the duration of its retention period. Protection comes from storage-level mechanisms such as WORM storage or S3 Object Lock. Even if an administrator account is compromised, the backup cannot be changed before the lock expires.

What it means for a business: Ransomware operators try to delete backups before encrypting production data. An immutable backup is one of the most effective ways to make sure a clean copy survives an attack.

Related: immutable backup storage · what is immutable backup · Object Lock

An incremental backup copies only the data that has changed since the previous backup of any type, whether that was a full or another incremental. A restore requires the last full backup plus every incremental taken after it, as an unbroken chain.

What it means for a business: Incremental backups are fast and small, so most daily backup jobs use them. Because one damaged link in the chain can affect later restore points, backup verification becomes important.

Related: differential backup · forever forward incremental · restore testing

Instant recovery starts a virtual machine directly from its backup file instead of waiting for a full restore. The machine runs from the backup repository first and is then moved to production storage in the background. Software such as Veeam offers this at the virtual machine level.

What it means for a business: While restoring a large server can take hours, instant recovery can bring the service back within minutes. The performance of the backup repository decides the user experience while the machine runs from it.

Related: RTO · backup and disaster recovery

Microsoft 365 backup keeps a separate copy of Exchange Online mail, OneDrive files, SharePoint sites and Teams data outside Microsoft’s own infrastructure. Microsoft is responsible for service availability; under the shared responsibility model, protection against data deletion and misuse largely falls to the customer.

What it means for a business: The recycle bin and retention policies offer time-limited protection and do not replace an independent backup. A separate backup becomes critical when a departing employee’s account is deleted or an account is compromised.

Related: Microsoft 365 backup · KVKK and data backup · SaaS

Object Lock is a WORM feature of S3-compatible object storage that prevents objects from being deleted or overwritten for a set period or indefinitely. According to AWS documentation, it works on buckets with versioning enabled and offers two retention modes, governance and compliance. In compliance mode no user can shorten the lock period.

What it means for a business: Backup software such as Veeam can write to Object Lock-enabled storage to create immutable backups. The mode and period chosen determine both the level of protection and the cost of data that cannot be deleted.

Related: S3 object storage · immutable backup · S3-compatible storage

Replication keeps a copy of a system or data set in another location up to date, either continuously or at set intervals. It can run at the virtual machine, storage or database level. The target copy stands ready to be brought online quickly when needed.

What it means for a business: Replication is used to shorten downtime, while backup is used to go back in time. Because accidental deletion or encryption at the source can also be copied to the target, replication complements backup rather than replacing it.

Related: data replication · synchronous replication · asynchronous replication

A restore point is a backup record that allows a system or data set to be returned to its state at a specific moment. Every successful backup job creates at least one restore point. The retention policy decides how many are kept.

What it means for a business: How often restore points are created sets the upper limit of potential data loss, in other words the RPO actually achieved. A system backed up once a day accepts roughly a day of data loss in the worst case.

Related: RPO · retention · backup and disaster recovery

Restore testing is a planned exercise to confirm that backups can actually be opened and that systems can be brought back within the target time. It can range from restoring a single file to starting a whole system in an isolated environment. Results and measured times are recorded.

What it means for a business: A backup job showing “success” does not prove the data can be recovered. Regular testing reveals the real RTO and is the basis for the “0 errors” requirement in the 3-2-1-1-0 rule.

Related: 3-2-1-1-0 rule · backup management · disaster recovery plan

Retention is the setting that defines how long, or how many, backups or restore points are kept before deletion. It can be expressed in days, weeks or a number of points. Backups that pass the limit are deleted or merged automatically by policy.

What it means for a business: Retention sets both the oldest date you can return to and the storage cost. Since ransomware often goes unnoticed for weeks, a very short retention period can make it impossible to reach a clean backup.

Related: GFS retention · KVKK and data backup · backup storage calculator

RPO (Recovery Point Objective) is the point in time to which data must be recovered after an outage. NIST SP 800-34 defines the term in exactly these words. In practice it expresses the maximum acceptable data loss as a period of time.

What it means for a business: An RPO of 4 hours means the business accepts losing the orders, invoices or records entered in the last 4 hours. Backup frequency and replication method are chosen to meet this target.

Related: RTO · disaster recovery · disaster recovery plan

RTO (Recovery Time Objective) is the maximum time a system can remain in the recovery phase after an outage before business processes are negatively affected. It sets the target time from the start of an outage until the service is usable again. The definition comes from NIST SP 800-34.

What it means for a business: The shorter the RTO, the more expensive the solution: an RTO of minutes may require a standby replica, while an RTO of days may be met by restoring from backup. The right value is set together with the hourly cost of downtime.

Related: RPO · downtime cost calculator · disaster recovery

A snapshot is a record of the state of a virtual machine, disk or storage volume at a specific moment. It usually lives on the same storage as the source system and is created within seconds. Changes are tracked in a separate layer, so it uses very little space at first.

What it means for a business: Snapshots are handy for a quick rollback before an update, but on their own they are not a backup: if the storage fails, the snapshot is lost too. Long-lived snapshots can also slow performance, so lasting protection comes from a separate backup.

Related: backup · 3-2-1 rule · cloud backup

Synchronous replication is a type of replication in which a write is not acknowledged to the application until it has been completed at both the source and the target. The two copies are always in the same state and theoretical data loss is close to zero. Because every write waits for the target, distance and network latency directly affect performance.

What it means for a business: Synchronous replication suits systems that cannot tolerate data loss, but it is usually practical only between nearby, low-latency sites. Bandwidth and infrastructure costs are higher than with the asynchronous method.

Related: asynchronous replication · RPO · data replication

A synthetic full backup is a full backup built by combining existing full and incremental backups in the repository, without reading data from the production system again. The result has the same content as a conventional full backup taken at the same time. The whole process runs inside the backup infrastructure.

What it means for a business: Synthetic fulls shorten long incremental chains and spare production servers from heavy full-backup reads. For companies with a narrow backup window, they are a practical way to keep regular full backups.

Related: full backup · forever forward incremental · cloud backup

WORM (Write Once, Read Many) is a storage model in which data, once written, cannot be changed or deleted for a set period but can be read as often as needed. It was once implemented with optical discs and special tapes; today it is mostly done with software locks in object storage.

What it means for a business: WORM is the technical basis both for immutable backups and for records that regulations require to be kept unaltered. Because a locked retention period may not be shortened later, it has to be set correctly.

Related: immutable backup · Object Lock · S3 object storage