Data Protection and Continuity

What Are RTO and RPO? How to Set Business Continuity Objectives

Nanda Teknoloji Engineering Team · · 9 min read

In brief

RTO states how soon a system must be back up after an outage at the latest; RPO states how much data loss can be tolerated at most. Both are business decisions, not IT decisions, and are set separately for each system.

RTO (Recovery Time Objective) is the time target that shows how soon, at the latest, a service must be usable again when a system goes down. RPO (Recovery Point Objective) is the data target that shows how much data, counting backward from the moment of the outage, can acceptably be lost. One answers “how long can we wait?” and the other answers “how much can we lose?” These two numbers determine which backup frequency, which replication model, and which disaster recovery architecture you need. They are therefore not a technical detail but the backbone of the business continuity plan. In this article we walk step by step through the two concepts, the difference between them, how to derive objectives with a business impact analysis, and how to translate objectives into architecture.

Seeing RTO and RPO on a timeline

The easiest way to think about the concepts is on a timeline. The moment of the outage sits in the middle. To the left is the last usable backup or replica point; to the right is the moment the service is back in operation.

  • The distance between the outage and the last valid data point is the data loss actually experienced. RPO is the upper limit of this distance.
  • The distance between the outage and the moment the service returns is the downtime actually experienced. RTO is the upper limit of this distance.

For example, if you set the RPO for an order system at “15 minutes,” the backup or replica point must never be older than 15 minutes. A backup taken once a night does not meet this target. If you wrote an RTO of “four hours,” you need to prove that the servers, network, application, and data can be brought to a working state within four hours.

For short definitions, see the glossary entries for RTO and RPO.

Common confusions about RTO and RPO

The two concepts are often confused. The table below summarizes the difference.

Feature RTO RPO
What it measures How long the service stays down The time span of lost data
Unit Minutes, hours, days Minutes, hours, days
Defining question How long can we wait? How much can we recreate?
Technical decision it drives Speed of the recovery process, spare capacity, automation Backup frequency, replication type
Typical mistake Counting only the data restore in the time Confusing the moment a backup is taken with the moment it is consistent

One more distinction is needed: RTO and RPO are objectives, not guarantees. The SLA commitments in a service provider’s contract and your internal objectives are not the same thing. The objective is your business requirement; the architecture and the contract are built to meet it.

How are RTO and RPO set? Business impact analysis

The right objective comes not from a server inventory but from a business impact analysis (BIA). NIST’s SP 800-34 Contingency Planning Guide also builds planning on the system inventory and the business impact analysis. In practice, these steps work:

  1. List the business processes. Start with processes, not systems: order intake, invoicing, payroll, customer support, production planning.
  2. Map each process’s dependencies. Which application, database, authentication service, network component, and external provider does it rely on?
  3. Ask about the impact of an outage. What happens if the process stops for an hour? For a day? Get qualitative and, where possible, numerical estimates under headings such as revenue loss, legal obligations, reputation, and employee productivity.
  4. Ask about the impact of data loss separately. Is it possible to re-enter the last hour of orders? Is there data that cannot be recovered from an external source, such as a bank receipt?
  5. Create criticality tiers. Divide systems into three or four tiers and assign an RTO/RPO pair to each tier.
  6. Get management approval. Objectives must be signed off by business owners and senior management; IT only reports their cost and feasibility.

The output of this work should be a table. The example below shows the tier logic for a fictional mid-sized company. The numbers are for illustration only; for your own organization they must come from your business impact analysis.

Tier Example systems Example RTO Example RPO Typical architecture
1 - Critical Orders, payments, authentication Minutes Near zero Synchronous or near-synchronous replication, a ready environment at a second location
2 - High ERP, CRM, email A few hours A few tens of minutes Frequent replication, a recovery environment on standby
3 - Medium Reporting, internal portal One business day One day Daily backup, recovery on demand
4 - Low Archive, test environments Days One week Low-frequency backup

How do objectives turn into architecture?

Once the objectives are set, the next step is choosing the technical structure that will meet them. Three rules help here.

RPO determines backup and replication frequency. You cannot bring RPO down to a few hours with a daily backup. You need more frequent backups, continuous data protection, or replication. But replication alone is not a backup: corrupted or encrypted data is copied just as quickly. That is why replication is considered together with a point-in-time, immutable backup. We covered this topic in detail in our article on the 3-2-1-1-0 backup rule.

RTO determines the recovery method. Restoring data from backup is often the slowest step. As RTO gets shorter, the need grows for infrastructure on standby, automatic failover, and rehearsed runbooks. If you are willing to accept a long RTO, building from scratch from backup may be sufficient and cheap. We compare the options in between in the article on the disaster recovery center and DRaaS.

Cost rises exponentially, not linearly, as objectives tighten. The difference between “minutes” and “hours” is far more expensive than the difference between “hours” and “days.” Showing business owners the options from this angle is an effective way to have a realistic conversation about objectives.

Common mistakes

  • Covering all systems with a single RTO/RPO. Either you overspend or you leave the critical system underprotected.
  • IT setting the objective alone. An objective without business owner approval becomes a point of dispute in a crisis.
  • Skipping dependencies. If the application comes up in 2 hours but the authentication service or DNS comes up in 6 hours, the real RTO is 6 hours.
  • Confusing “a backup was taken” with “it can be restored.” An untested backup does not prove your RPO. For the test method, see the disaster recovery test checklist.
  • Not counting post-recovery steps. After the data is back, application validation, user access, network routing, and integration checks are also part of the RTO.
  • Writing objectives once and leaving them. As architecture and the business change, objectives go stale.

Measuring and proving objectives

Writing an objective is not enough; it must be measurable. In every exercise, record the actual recovery time (achieved RTO) and the data point you returned to (achieved RPO) and compare them with the objective. If there is a gap, one of two paths is chosen: improve the architecture or have the objective realistically re-approved. The records are also valuable for audits and for business continuity governance.

Example scenario: One order system, two different objectives

Say an e-commerce company’s order system consists of two parts: the order database and the content repository where product images are kept. The business impact analysis yields the following: orders cannot be recovered from outside, and because the customer’s card charge has already been completed, a lost order directly means reputation damage and a refund burden. The images, on the other hand, can be re-uploaded from another source, and missing them for a few hours does not block an order.

As a result, a short RPO and short RTO are set for the database, and relaxed objectives for the content repository. Two components of the same application fall into different tiers. This is a typical example of why the “single objective per application” approach is misleading. Thinking about objectives at the component level directs spending to where it really matters.

Questions to ask in an objective-setting meeting

Conversations with business owners often start with the answer “there should be no outage at all.” You can use these questions to make that answer concrete:

  • When this system goes down, which business process stops and who is affected?
  • Is there a difference between the first hour and the eighth hour of an outage? When does the impact rise sharply?
  • Can we make up for data loss manually? How many people and how much time would that take?
  • Is there a legal, contractual, or audit-bound time limit?
  • While this system is down, is a workaround (a manual process, an alternative tool) possible?
  • Are there times, such as period end, campaigns, or payroll, when an outage is much more expensive?

The last question is especially important: objectives do not have to stay fixed throughout the year. Some organizations temporarily apply tighter protection during critical periods.

Factors outside scope

RTO and RPO are meaningful not only in disaster scenarios but also in planned maintenance and minor failures; however, trying to manage all causes of outage with the same objective is misleading. A single disk failure, the loss of a data center, and a ransomware attack call for different technical responses. That is why, when writing objectives, you should also state which scenarios are covered: the objective met by a high availability cluster structure and the objective for a site disaster can be written separately. The availability percentage does not replace these objectives; a monthly availability percentage does not tell you how long a single outage will last or how much data will be lost.

Decision summary

  • Group systems into tiers by business impact; assign a single RTO/RPO pair to each tier.
  • Match RPO to backup and replication frequency, and RTO to the recovery method.
  • Approve objectives together with business owners and show their cost transparently.
  • Measure the achieved values in every exercise; close the gap with the objective.
  • Review objectives at least once a year and after major changes.

Moving forward with Nanda Teknoloji

On projects like this, Nanda Teknoloji carries out the discovery, design, migration, and operations steps: validating the objectives that come out of the business impact analysis, designing the backup and recovery architecture that meets them, and proving it with regular exercises. For our related solutions, see the backup solutions and disaster recovery center pages.

Frequently asked questions

What is the difference between RTO and RPO?

RTO measures time, meaning how quickly the service must come back. RPO measures data, meaning how much data, counting backward from the moment of the outage, can be lost.

Can RTO and RPO be zero?

In theory they can be written down as targets, but as you approach zero, cost and complexity rise quickly because synchronous replication, multiple locations, and spare capacity are required. It is more sensible to choose a realistic target for most systems.

Can the same RTO and RPO be set for all systems?

It is not recommended. A payment system and an internal document archive have different outage tolerances. Grouping systems into criticality tiers and assigning separate objectives to each tier is both cheaper and easier to defend.

How often should RTO and RPO objectives be reviewed?

At least once a year and after major changes: a new application, an architecture change, a merger, or a new legal obligation. Objectives should also be compared against test results.

Related solutions

Related posts

Data Protection and Continuity

How to Run a Disaster Recovery Test: A Checklist

Disaster recovery test types and the steps for preparation, execution, and evaluation, with a practical DR test checklist you can copy and use.

· 8 min read

Let's plan your critical systems together

Tell us what you need; our team will work out the right architecture and migration path with you.