Defining RTO and RPO for an Effective Disaster Recovery Plan

Defining RTO and RPO for an Effective Disaster Recovery Plan

An outage lasting a few hours may be tolerable for a support function, but could become critical for a trading floor, a payment platform, or a healthcare service. Defining RTO and RPO involves precisely translating this operational reality into measurable objectives for recovery and data restoration. These two metrics are neither technical default values nor mere compliance requirements: they represent governance decisions that involve business units, IT, cybersecurity, and service providers.

Defining RTO and RPO: Two Complementary Objectives

RTO, which stands for Recovery Time Objective, refers to the maximum acceptable duration of downtime for a business process, application, or service following an incident. It answers the question: How long can the organization operate without this capability before the consequences become unacceptable?

A four-hour RTO means that the service must be restored to operational status within four hours of an outage, depending on the defined scope. This timeframe generally includes incident detection, the decision to activate the RTO, mobilizing teams, any necessary failover, restoration operations, testing, and restoring service availability. A realistic goal is therefore never limited to the time it takes to restart a server.

RPO, which stands for Recovery Point Objective, refers to the maximum acceptable data loss, expressed in terms of time. It answers another question: How far back in time is the organization willing to go after a restore? An RPO of fifteen minutes means that, in the worst-case scenario, data created or modified during the last fifteen minutes could be lost.

RTO and RPO are related, but they are not interchangeable. An application may need to resume within one hour while tolerating four hours of data loss. Conversely, a system may require virtually zero data loss but tolerate several hours of downtime. The first scenario primarily requires rapid recovery capabilities; the second requires, above all, more frequent replication, logging, or backup mechanisms.

Start with the business impacts, not the existing architecture

The most common mistake is to ask the IT department what RTOs and RPOs it can guarantee with the resources currently available, and then to treat that response as a business requirement. The process should follow the opposite approach. Recovery objectives are established based on the consequences of an outage and then evaluated against the technical, organizational, and financial capabilities required to achieve them.

The Business Impact Analysis(BIA) provides the framework for this decision. It helps identify priority activities, their dependencies, the minimum resources required for their operation, and how the impacts evolve over time. These impacts may be financial, regulatory, contractual, operational, reputational, or related to personal safety.

For an order-processing operation, a 30-minute outage can result in a direct loss of revenue, a breach of service commitments, and regulatory exposure. For internal administrative management, the same downtime will not necessarily have a significant impact. Therefore, these two activities should not receive the same level of investment or be subject to the same recovery objectives.

The BIA must also distinguish between business activities and their technical components. A critical business activity often depends on applications, data flows, directories, workstations, telecommunications, cloud providers, and reference data. A two-hour business RTO can only be met if all of these dependencies have been analyzed, prioritized, and incorporated into the recovery plan.

The maximum acceptable delay is not the RTO

The maximum acceptable downtime, sometimes referred to as MTPD or MTPOD depending on the standards, represents the point beyond which the impacts become unacceptable. The RTO must be below this threshold. It must allow for a margin to manage recovery contingencies, validate the return to service, and address the challenges specific to a real crisis.

Setting an RTO equal to the maximum tolerable delay means leaving no safety margin. In the event of a cyberattack, a service provider failure, or the simultaneous unavailability of multiple components, this approach exposes the organization to the risk of immediately exceeding its tolerance threshold.

How to Set Coherent and Achievable Goals

Defining RTOs and RPOs requires collaboration among business leaders, BCP/DRP managers, IT teams, information systems security, risk management, and—depending on the context—the legal and compliance functions. The goal is not to reach an abstract consensus, but to formalize trade-offs that are documented and approved by governance.

The first step is to categorize activities and services based on their criticality. This prioritization must be based on plausible disruption scenarios: site unavailability, infrastructure failure, data corruption, a ransomware attack, a telecommunications outage, or the unavailability of a critical supplier. An acceptable RTO during a local outage may become insufficient when the scenario involves a cyber compromise and longer remediation efforts.

The second step is to assess tolerance for data loss. A distinction must be made between transactional data, reference data, supporting documents, customer communications, and logs required for investigations. Backup frequency alone is not enough: one must also consider the time it takes to detect corruption, the ability to restore a healthy version, the integrity of replicated data, and business reconciliation processes.

The third step involves translating objectives into capacity requirements. A short RTO may require a preconfigured backup infrastructure, automated procedures, on-call teams, and regular testing. An RPO close to zero may require synchronous or near-synchronous replication, which can sometimes entail significant costs and performance constraints. These choices must be commensurate with the level of impact the organization is willing to accept.

Finally, decisions must be documented in a controlled set of records: BIA results, assumptions made, owners of objectives, dependencies, business continuity solutions, testing procedures, and validation criteria. RTOs and RPOs lose their value if they are not approved, tested, or revised as the business evolves.

Technical and financial trade-offs that need to be made transparent

A 15-minute RTO and a zero RPO are sometimes taken for granted, particularly after a major incident. However, these requirements can involve a considerable degree of complexity. They often entail redundant architectures, enhanced monitoring, proven failover procedures, protection against replicated data corruption, and personnel who can be mobilized at any time.

The expected level of service must be distinguished from mere nominal availability. A platform may be technically accessible without enabling teams to perform essential operations—for example, if data is incomplete, if a third-party data feed is unavailable, or if access permissions have not been restored. Resumption must therefore be validated based on business criteria, not solely on the technical status of the components.

Cybersecurity requires special attention. Rapid replication can spread malicious deletion or data corruption. For environments exposed to ransomware, RPO objectives must be supplemented by a strategy of isolated backups—immutable where appropriate—that are regularly restored to a test environment. The goal is not just to recover quickly—it is to recover a reliable, usable, and verified state.

Testing RTOs and RPOs under realistic conditions

An untested objective remains a hypothesis. Exercises must measure the actual time required—from the alert to business validation—and not just the time it takes to run a recovery script. They must also verify the actual recovery point achieved: what is the latest intact data recovered, which transactions need to be replayed, and who is responsible for reconciliation?

Tests benefit from incorporating realistic constraints: the unavailability of a key stakeholder, overloaded communication channels, reliance on a service provider, degraded remote access, or the need to preserve evidence for an investigation. Any discrepancies identified should inform a prioritized improvement plan, with designated responsible parties and deadlines.

This discipline is particularly useful during architectural changes, cloud migrations, service outsourcing, or the introduction of new regulatory requirements. Each transformation can alter dependencies and render previously coherent objectives obsolete.

Defining appropriate RTOs and RPOs essentially means explicitly stating the level of disruption that the organization is willing to accept or refuse. This decision requires a method, evidence, and professionals capable of linking business requirements to recovery capabilities. It is this integration of governance, impact analysis, and operational exercises that DRI France’s professional development programs aim to strengthen.

This post is also available in: French