RTO is the maximum downtime your business can accept, and RPO is the maximum data loss you can accept, measured as time back to your last good copy. They are independent numbers, so a system can meet one and fail the other. Your business, not your IT team alone, should set both through a business impact analysis.
TL;DR:
- Run a business impact analysis for each critical process, set its RTO below its maximum tolerable disruption period, and obtain business owner approval.
- Tier systems by business impact: revenue and safety systems may need recovery in minutes, while archives can tolerate days, subject to contractual and regulatory limits.
- Match technology to both targets: frequent backups reduce data loss, while warm or hot standby shortens downtime; synchronous replication costs more and has distance limits.
- Test failovers and restores under realistic conditions, record actual recovery times, and revise either the plan or target when drills expose gaps.
- For ransomware, keep offline or immutable copies with separate credentials, because replication can copy encrypted or deleted data before you can recover.
Table of Contents
- How do RTO and RPO differ side by side?
- How do you calculate and set RTO and RPO?
- How do cost, risk and priority shape your targets?
- Which technical approaches deliver which targets?
- How do you test and validate your RTO and RPO?
- What is a practical checklist for RTO and RPO?
- What do RTO and RPO look like in real examples?
- How do RTO and RPO shape your cybersecurity strategy?
- How do managed IT services and cloud solutions fit in?
- How do RTO and RPO guide incident response in small and medium businesses?
- Why does staff training decide whether you hit your targets?
- Our view: vendor-agnostic support for RTO and RPO
- How can we help you implement and test RTO and RPO?
- FAQ
- Sources
How do RTO and RPO differ side by side?
Think of RTO as time moving forward from the moment of failure, and RPO as time moving backward to the last valid backup or sync point. One answers “how long can we be down?” and the other answers “how much work can we afford to redo?”
- What it measures: RTO measures the length of an outage. RPO measures the age of the data you restore.
- Typical units: Both are expressed in minutes, hours or days. A payment system might need a very short RTO, while a file archive might tolerate a much longer one.
- What drives it: Backup and replication frequency drive RPO. Failover architecture, automation and staffing drive RTO.
- How one can fail while the other passes: Nightly backups give you an RPO of up to 24 hours, yet a warm standby might bring services back in 30 minutes. The reverse also happens: continuous replication keeps data fresh, but rebuilding servers by hand can take much longer.
- What failure looks like: A missed RTO means staff and customers sit idle. A missed RPO means re-keying orders, chasing lost emails or reconciling gaps in your accounts.
Set the two numbers separately for every critical process. Do not assume that fixing one fixes the other.
How do you calculate and set RTO and RPO?
Start with the business, not the technology. The steps below take you from process priorities to system targets.
- Run a business impact analysis (BIA). List your critical processes, such as invoicing, order entry and customer support, and map the systems and people each one depends on. NIST SP 800-34 provides a BIA template you can adapt.
- Find the maximum tolerable period of disruption (MTPD). This is the point where an outage becomes critical for the business. It is a business property, and your RTO must sit below it. Commvault gives a worked example: if impact becomes critical at 24 hours, the MTPD is 24 hours, and an RTO of 4 hours leaves a safety buffer.
- Decide how much data each process can lose. Ask finance, sales and operations how many hours of work they could realistically recreate. That answer becomes the process-level RPO.
- Translate process targets into system targets. If order entry needs a short RTO, the database, the network and the identity service behind it all need an RTO at least as short.
- Document assumptions and get sign-off. Write down what each target assumes, such as staff availability or internet access, and have business owners approve the numbers. For a wider plan around these targets, see our guide to a business continuity plan for SMEs.
How do cost, risk and priority shape your targets?
Tighter targets cost more, and the relationship is steep. Moving from a longer RTO to a very short one usually means duplicate infrastructure, automated failover and more skilled staff. Moving from a 24-hour RPO to near-zero means replication running all day, every day.
That is why you tier systems rather than apply one target to everything:
- Tier 1, minutes: Systems that stop revenue or put safety at risk, such as payment processing, core line-of-business applications and telephony.
- Tier 2, hours: Systems that hurt productivity but allow workarounds, such as email, file shares and internal reporting.
- Tier 3, days: Archives, development environments and rarely used tools.
- Regulatory and contractual limits: Customer contracts, insurance conditions or industry rules may set a minimum, so check them before you finalise any tier.
- Trade-off conversations: Show stakeholders the cost of each tier alongside the business impact of downtime. If the price of a very short RTO exceeds the loss from a longer outage, a looser target is the sensible choice.
Review the tiers once a year, or whenever a system changes its role in the business.
Which technical approaches deliver which targets?
Once targets are agreed, map them to the patterns that can meet them.
- Backup frequency and retention: Your backup cadence sets your RPO. Less frequent backups can mean losing more data, while more frequent snapshots reduce that potential loss. Our backup retention policy guide covers how long to keep each copy.
- Synchronous replication: Every write lands at two sites before it is confirmed, which gives near-zero RPO at higher cost and with distance limits.
- Asynchronous replication: Data is copied with a short lag, which saves money but accepts a small loss window.
- Hot standby: A running duplicate environment can take over quickly.
- Warm standby: Systems are built but scaled down, so failover takes longer.
- Cold standby: Infrastructure is rebuilt from backups after a disaster, which can take much longer.
- Continuous data protection and distributed architectures: These capture every change or spread load across locations, giving low RPO and low RTO together.
Pro Tip: Measure real failover times in a test, because a vendor’s service level agreement tells you what they promise, not how your environment behaves under load.
How do you test and validate your RTO and RPO?
A target you have never tested is a guess. Yale’s security standard states the principle plainly: if the disaster recovery plan cannot deliver recovery within the defined RTO, then either the plan or the availability requirement must change.
- Schedule regular failover drills. Exercise both how long recovery takes and whether the restored data is complete and usable. IBM’s guidance on DR simulation describes simulating emergency conditions and restoring data to verify runbooks and targets.
- Avoid the common antipatterns. Untested failovers, stale runbooks and neglected standby sites fail when you need them most.
- Automate where it is safe. AWS notes that Elastic Disaster Recovery can run drill instances without redirecting production traffic, which cuts manual error and risk.
- Record, adjust and re-test. Log actual recovery times, fix what broke, and change either the plan or the target. Cutover argues that recovery plans go stale quickly, so treat testing as continuous operational work rather than a one-off project.
What is a practical checklist for RTO and RPO?
Use this as your next-steps list.
- Complete a business impact analysis and document the MTPD for each critical process.
- Set RTO and RPO per process, then get written sign-off from the business owners.
- Choose the technical measures that match each tier and write the runbooks for them.
- Schedule recovery drills, assign a named owner to each, and put the dates in the calendar.
- Review the results, update the targets or the plan, and repeat the cycle regularly.
If you want a template to turn this list into a full plan, our guide to building an IT disaster recovery plan is a good next read.
What do RTO and RPO look like in real examples?
Concrete numbers make the two metrics easier to hold apart. The figures below are illustrative, not market data.
An accounting practice. The practice runs its practice-management database on a server backed up at 11 pm each night. If the server fails at 3 pm, the last good copy is 16 hours old, so up to 16 hours of time entries and invoices are gone. That is the RPO problem. If a new server takes a long time to build and restore, that is the RTO problem. The practice might set a short RPO and RTO for that system, then move to more frequent snapshots and a prebuilt standby.
A retailer with an online shop. Say the shop takes orders around the clock. A very short RTO and an RPO near zero may make sense, because every lost order is lost revenue and a customer who may not return. The same business can accept a much longer RTO for its staff handbook archive.
A small professional services firm. Email and shared documents sit in the cloud, so the vendor keeps the service running, but a deleted mailbox or a corrupted shared folder is still your problem. A realistic target is an RPO of 24 hours and an RTO of four hours for file restores. For the cloud side of this, see our notes on Microsoft 365 backup and restore testing.
Notice that in each case the two numbers differ. That is normal, and it is the reason to set them separately.

How do RTO and RPO shape your cybersecurity strategy?
Ransomware changes the maths. Replication that copies data instantly also copies encrypted or deleted files instantly, so a near-zero RPO from replication alone can mean near-zero protection against an attack. You need backup copies that an attacker cannot reach or alter, kept apart from your production network.
That is what a ransomware-safe backup means in practice: offline or immutable copies, separate credentials, and restore points that predate the infection. Your RPO then has to account for how far back you may need to go to find a clean copy, which can be days rather than minutes.
Your RTO matters just as much. After an attack you may need to rebuild systems from scratch, rotate every password and confirm the intruder is gone before you reconnect anything. That takes longer than a hardware failover, so a cyber-specific RTO is often longer than your standard one. Plan for it openly rather than discover it mid-incident.

The FBI’s Internet Crime Complaint Center annual report documents cyber incident trends and reinforces why tested backups and recoveries matter for every organisation. Our cybersecurity service builds backup and recovery into a wider defence, so your targets hold up when the cause of the outage is hostile.
How do managed IT services and cloud solutions fit in?
Handing infrastructure to a managed provider or a cloud platform does not hand over your RTO and RPO. It moves where the work happens. A cloud vendor may keep its platform available, yet you remain responsible for protecting your own data inside it, including accidental deletion and account compromise.
So write your targets into the service arrangement. A good managed service agreement states which systems are covered, how often they are backed up, how restores are tested, and how fast the provider responds when something fails. If those points are vague, your targets are only hopes.
Cloud makes some patterns cheaper. You can keep a warm standby environment scaled down until needed, or replicate to another region without buying hardware. You also gain monitoring that flags a failed backup job the same day rather than at the next restore attempt, which protects your RPO quietly in the background.
Ask your provider practical questions. When did you last restore this system, and how long did it take? Where do the backup copies live, and can an attacker with our admin credentials delete them? Who calls whom when the failover starts? Clear answers mean the targets are real.
Our managed IT services and Cloud and Microsoft 365 support are set up around exactly these questions, and our business IT support checklist shows the operational steps that keep downtime short.
How do RTO and RPO guide incident response in small and medium businesses?
Smaller businesses rarely have a dedicated recovery team, so the targets do double duty: they set the order of work during an incident. When everything is down, the tier list tells you what to restore first, and the RTO tells you how long you have before the damage compounds.
Keep the plan short enough to use under stress. A concise response sheet should list who declares an incident, who calls the IT provider, which systems come back first, and what you tell customers and staff while you work. Pair each critical system with its RTO and RPO so nobody debates priorities in the middle of a crisis.
Budget is the other reality. You probably cannot afford hot standby for everything, and you do not need it. Spend where the business impact is highest, accept longer targets elsewhere, and write down the decision so the next manager understands why. A cheaper tier that is tested beats an expensive one that is not.
Include your communications in the plan too. Customers forgive an outage that is explained clearly and resolved in the time you promised. They do not forgive silence, or an estimate that turns out to be fiction. An honest RTO, backed by drills, lets you give a number you can keep.
Finally, link the incident plan to your wider continuity work. A recovery target for email is of little use if staff have no way to reach each other while it is down.
Why does staff training decide whether you hit your targets?
The best architecture still depends on people. In a real incident, someone has to recognise the problem, escalate it, follow the runbook and make decisions under pressure. If those people have never rehearsed, your actual RTO will be far longer than the one on paper.
Train the relevant groups. General staff need to know how to spot and report suspicious activity quickly, because early reporting shortens recovery. Your technical team needs hands-on practice with the restore and failover steps, ideally in the drills described above. Managers need to know their role: approving priorities, speaking to customers and authorising spend.
Make the training concrete. Run a tabletop exercise where you walk through a ransomware scenario and ask who does what at each stage. Time it. Gaps show up quickly, such as a runbook that assumes access to a system that is itself down, or an on-call contact who left the business last year.
Refresh the training whenever people or systems change, and after every drill. Short, regular sessions stick better than an annual marathon. Our cybersecurity training for staff covers the awareness side, which is often where an incident starts and where the first minutes of your RTO are won or lost.
Our view: vendor-agnostic support for RTO and RPO
We discuss backup practices with businesses. We help businesses establish trusted recovery-time targets. We help with the business impact analysis, the backup strategy and the recovery tests, and we give vendor-agnostic advice so your resilience matches your budget rather than a product catalogue. If you lack the capacity to test, a short assessment is a sensible first step.
— Ru
How can we help you implement and test RTO and RPO?
Our managed IT and disaster recovery support turns your targets into something you can rely on, backed by an emergency response team and advice that is not tied to any single vendor.

- Business impact analysis assistance to set realistic targets
- Backup design and ransomware-safe copies
- Recovery testing with documented results
- Emergency response when something fails
Request an assessment or book a recovery test through our managed IT services page.
FAQ
What is the difference between RTO and RPO?
RTO is how long you can be down before the damage becomes unacceptable. RPO is how much data, measured in time, you can afford to lose. They are independent, so you set each one separately for every critical system.
What is RTO and RPO and MTPD?
MTPD is the maximum tolerable period of disruption, the point where an outage becomes critical for the business. Your RTO should be shorter than the MTPD to leave a safety margin, while RPO limits how far back your data can be restored. Commvault’s example sets a 4-hour RTO against a 24-hour MTPD.
Can you give me an example of an RTO?
A very short RTO and an RPO near zero may make sense for an online shop’s order system, because every lost order is lost revenue and a customer who may not return. The same business can accept a much longer RTO for its staff handbook archive. The right figure depends on the business impact, not on what technology can do.
Can RPO be longer than RTO?
Yes. The two are unrelated measures, so an RPO of 24 hours alongside an RTO of 2 hours is entirely possible. It would mean you restore service quickly but accept losing up to a day of data.
Sources
- Commvault
- Cutover
- YALE-MSS-3.2: Test the Disaster Recovery Plan
- REL13-BP03 Test disaster recovery implementation to validate the implementation – Reliability Pillar
- Disaster recovery testing
