How to Set and Approve MTPD, RTO and RPO Targets

Reconciling MTPD, RTO, RPO and MBCO targets between business demand and the technical capacity measured in testing: a sample calculation, gap table, decision tree and management approval.
Technical summary
Technology
ISO 22301, ISO 22300, NIST SP 800-34
Topic
Setting and approving continuity targets
Level
Intermediate–advanced
Prerequisites
  • A completed business impact analysis
  • The latest restore test results
In this article7 sections

The business unit says, “The system must never stop, we must never lose data.” IT knows the current infrastructure can only recover in a day, from last night’s backup. Both sides repeat their sentence in the meeting, nobody decides, and in a real outage both the target and the capacity stay on paper. MTPD, RTO and RPO targets are where these two realities are reconciled by a management decision. Below we explain how to translate the concepts into management language, how to put the gap between demand and capacity into a table and how to get targets approved and measured.

Concepts: MTPD, RTO, RPO, MBCO

The terms are defined in the ISO 22300 vocabulary; the BIA clause of ISO 22301:2019 (8.2.2) mentions MTPD and RTO in its notes. NIST SP 800-34 uses a similar concept under the name MTD (maximum tolerable downtime). In a management presentation, each concept must have a single meaning; otherwise the meeting turns into a debate about definitions.

Concept The question it asks Whose decision
MTPD (maximum tolerable period of disruption) How long can this activity stop at most before the impact becomes unacceptable? Top management, based on BIA data
RTO (recovery time objective) By when must the activity resume at a minimum level? The process owner proposes, management approves
RPO (recovery point objective) Going back from the moment of disruption, how much data loss at most is acceptable? The process and data owner propose, management approves
MBCO (minimum business continuity objective) What minimum service level is acceptable during the disruption? Process owner and management

In most organizations, MBCO is left undefined, and that’s why the RTO discussion gets blurred. The sentence “The order process must be back in 8 hours” is incomplete. The correct version is something like: “The order process must be back within 8 hours, at a level that can process the orders of the most critical customer group at part of the normal daily volume.” Once the minimum level is defined, the recovery scope narrows too; perhaps not the whole ERP but only the order module and the linked stock query are enough. That means a smaller DR environment.

The logical relationship between targets

Targets aren’t independent numbers. Catch inconsistencies before approval; if you catch them after, you’ll have to withdraw a signed document. The basic rules:

  • RTO < MTPD: If RTO is set equal to MTPD, the smallest delay turns directly into an unacceptable impact. The margin in between is for noticing the incident, making a decision and verifying after recovery.
  • Dependency order: An activity’s RTO can’t be shorter than the RTOs of the activities feeding it and the systems it uses. That’s why infrastructure components such as authentication, the network and DNS carry the shortest RTO. No application signs in until the domain controller is up.
  • RPO is measured in time but isn’t a duration: RPO doesn’t define how long recovery will take but which point in time the data will return to. A short RTO with a long RPO (coming back fast but with old data), or the reverse, is possible.
  • Re-entering lost data is part of the RTO: If the RPO is four hours, the lost four hours of transactions need to be re-entered from paper or email. This time is added on top of technical recovery.

A sample calculation: suppose an activity’s MTPD is 48 hours. If noticing the incident and escalation take 2 hours, the crisis team’s decision to declare a disaster 2 hours, technical restore 12 hours, and application verification and re-entering lost data 6 hours, the total is 22 hours. In this case, a 24-hour RTO is both realistic and leaves a margin against the MTPD. If the business unit wants a 4-hour RTO for the same activity, technical restore time alone exceeds the target. The gap won’t close without investment.

The gap between business demand and technical capacity

Targets from the BIA are the demand side; how they’re gathered is explained in the business impact analysis guide. The capacity side is the recovery time and data loss IT can prove today. There are two traps we keep running into in the field:

  1. Capacity on paper: The sentence “We have replication, we’ll be up in 15 minutes” isn’t capacity if it hasn’t been tested. When the replica is brought up at the target, the IP address may have changed, the DNS record may still point to the old address, and the application license may turn out to be tied to a MAC address. Write in the capacity column only the value measured in the last restore test; if it hasn’t been measured, write “unverified”.
  2. Looking at a single system: An application’s virtual machine may come up in an hour; but if the application depends on a database server, license server, file share and external integration, capacity is the time it takes for the whole chain to come up. The slowest link decides.

Which target is met with which technology (restore from backup, replica, cluster) is a separate engineering topic; for that comparison, see the article choosing technology based on RPO/RTO targets. The topic here is how the cost of those options is weighed against business demand at the management table.

Gap table and decision tree

The gap table should be the single document management decides on. A twenty-page DR report doesn’t get read; a four-column table does. A row is opened for each prioritized activity:

Activity Demand (RTO / RPO) Capacity measured in testing Gap and options
Order intake 8 hours / 1 hour 24 hours / 24 hours (last test) Replica + frequent log backups; or relax the target
Shipment 24 hours / 4 hours 24 hours / 24 hours RPO gap; manual delivery note workaround
Payroll 3 days / 24 hours 24 hours / 24 hours No gap

The values in the table are examples. For each gap, follow this decision tree:

  1. Was the demand really derived from the MTPD? If the business unit can’t justify the target with the BIA impact scale, the target is reassessed. Unjustified “zero downtime” requests are eliminated at this step.
  2. Is there a workaround? If the activity can be kept going at a minimum level (MBCO) through manual working, an alternative channel or supplier support, the system RTO can be extended. The workaround is written into the plan and tried in an exercise.
  3. What does closing the gap cost? The investment and operating cost of the technical options is compared with the time-based impact in the BIA. If the cost is proportionate to the impact, the investment goes into the budget plan.
  4. If it won’t be closed, is the risk accepted? This is a risk acceptance decision given with the signature of an authorized manager. Authority limits must be consistent with the organization’s risk acceptance criteria.
  5. Interim period: Even if the investment is approved, the real capacity that applies until it goes live is written into the plan. The plan can’t rely on a storage unit that hasn’t been bought yet.

Management approval of the targets

Approval has three layers. The process owner signs the target and justification for their activity. IT or the service provider signs the capacity column and the options for closing the gap. Top management makes the investment, workaround or risk acceptance decision for the whole gap table. The decision record contains:

  • Approved MTPD, RTO, RPO and MBCO values (per activity and supporting system)
  • Accepted gaps, the justification, validity period and the person accepting them
  • Approved investments, owner and target date
  • The next review date

This record becomes a direct input to the business continuity management system’s management review meeting; for the meeting’s input and output structure, see the management review guide. Approved targets must be carried over unchanged into the disaster recovery plan and service contracts (outsourced IT, cloud, data center). If the contract says “response within 48 hours at the latest” while the approved RTO is 8 hours, that’s a gap too. It’s a gap that’s easy to miss.

Periodic validation of targets

From a business continuity compliance perspective, an approved target is an assumption until it’s tested. Validation answers two questions: does the target still reflect the business requirement, and does capacity still meet the target?

The first question is answered by the BIA review. The second by measurement: the recovery time and data loss actually achieved in restore tests and DR exercises are recorded and compared with the approved target. Define the start and end of the measurement in advance. As RTO is counted from the disruption, the start is “the moment the disruption began” and the end is “the moment the process owner performs a test transaction in the application and gives approval”. The time of the disaster declaration decision is recorded separately; that way, detection and decision delays are seen separately from technical recovery time. That’s also why the sample calculation above includes detection and decision times. Measuring only how long the virtual machine takes to start doesn’t show the real RTO on the user side. A ping response isn’t recovery.

Which types of exercise suit which validation is compared in the article types of business continuity exercises. If a test result doesn’t meet the target, it’s recorded as a corrective action and the gap table is updated. To link targets to the disaster recovery plan and recovery procedures, see the disaster recovery plan (DRP) service page; to run the target work together with the BIA, see the BIA and BCP service page. For the official definitions of the concepts in the context of IT recovery planning, NIST SP 800-34 Rev. 1 is a good reference.

Frequently asked questions

Can RTO be set to the same value as MTPD?

Not recommended. MTPD is the point at which the impact becomes unacceptable; RTO is set before this point, leaving room for incident detection, decision-making and verification. If the values are equal, any delay directly creates an unacceptable impact.

Who sets the RPO: IT or the business unit?

The business unit and data owner propose the acceptable data loss, because they bear the consequences of the loss. IT shows which technical option can meet this request and at what cost. The final value is confirmed by management approval.

What should be done if the requested target can’t be met?

It’s written into the gap table and management is given three options: increase capacity through investment, maintain the minimum service level with a workaround, or accept the risk with an authorized signature. Quietly changing the target to fit the capacity isn’t right.

When should targets be reapproved?

Once a year, together with the BIA review; and also when there’s a significant change to a process, system or service provider. If an exercise or a real outage shows the target doesn’t hold, the gap table is updated and resubmitted for approval.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Compliance

We work with you on compliance with ISO 27001, IATF 16949, TISAX and similar standards, covering risk assessment, control design, documentation and audit preparation.

Knowledge Center

Other articles