GUIDE · BACKUP DESIGN

Business backup strategy: how to design one from scratch

After this guide you will know where to start when designing a backup strategy: which systems to protect, which recovery objectives to set, how to organise copies and how to verify that they really work. It will help you decide on clear criteria and make a precise request to your supplier.

Updated october 2026
10 min read
Altovar Technical Editorial Team

1. Why start with the strategy, not the software

Many businesses buy a backup product before deciding what they must be able to get back. When a failure or serious error arrives, they discover that a copy existed but did not cover the right systems, or that restoring takes days instead of hours. A strategy is the written answer to three questions: which data and systems must become available again, within how long and with what maximum acceptable loss of work. The software comes after, and the choice depends on those answers.

A written strategy also serves other purposes. It justifies the spending to management, gives your supplier a criterion for proposing solutions and provides a basis for the checks required by GDPR, which requires the ability to restore availability of and access to personal data in a timely manner. The document does not need to be long. It must be approved by the owners, read by whoever responds in an emergency and updated whenever systems change.

2. Inventory: what really needs protecting

The starting point is an inventory of the systems holding valuable data: management system, email, shared files, application servers, databases, archives of cases or projects, accounting documents and customer data. For each one note the owner, where the data resides and which other systems depend on it. A well-made list takes a few hours and is more useful than a long project nobody keeps up to date.

Not all data carries the same weight. Some can be recreated from other sources; others are irreplaceable, such as an archive of documents that exists only on the server. Distinguish the two categories and give priority to the second. Also look for hidden systems: a service account, a shared folder on a department computer or a spreadsheet used for important decisions often appear in no protection plan, and they are precisely the ones nobody remembers having.

3. Recovery objectives: how much work and how much downtime

For each system two quantities must be set. The first is the maximum acceptable data loss, known as RPO (Recovery Point Objective): how much work you can afford to lose, measured as the interval between the last valid copy and the failure. The second is the maximum acceptable downtime, known as RTO (Recovery Time Objective): how long it can take before the system is back in service. Both are decided together with the department that uses the system, not by the IT department alone.

The objectives determine the frequency of copies, the architecture and therefore the cost. A system that can lose almost nothing needs very frequent copies or a replica; one that can stay down for longer can make do with daily copies. Write the objectives next to each inventory entry. If the supplier proposes a solution that does not meet them, the proposal is not adequate, whatever its price. The method for setting them is explained in the dedicated guide on RPO and RTO.

4. The 3-2-1 rule and separating copies

The 3-2-1 rule is a widely shared good practice and a useful starting point: three copies of the data, on two different media, with one copy off the main site. It is not a law, but it describes the problem well. A copy on the same server is lost together with the original during a fire, a theft or an attack that encrypts everything it finds on the network. Separation is the real goal, more than the exact number of copies.

Today the most frequent threat is ransomware, a malicious program that encrypts files and demands a ransom. For this reason the separate copy must be protected from modification and deletion even by whoever controls administrative accounts. Many systems offer an immutability option, meaning the ability to lock a copy for a period, or separation of credentials. Ask your supplier how the copy is protected, who can delete it and how it verifies this.

5. Backup, replication and snapshots: what each is for

Backup, replication and snapshots are different tools and not interchangeable. Backup creates historical copies that you can restore even weeks later, and protects against human errors and deletions. Replication keeps a nearly up-to-date copy on another system, useful for reducing downtime, but it also copies the errors. A snapshot is a quick picture of a volume's state, handy for rolling back a little, but it does not replace an independent copy of the system it protects.

The strategy combines these tools according to the objectives. A critical system may have replication for downtime and backup for data history; a document archive may be fine with daily copies. The detailed comparison is covered in the dedicated guide on backup, replication and snapshots. The principle here matters: no single tool covers all risks, and asking which risk each one covers is the right question to put to whoever proposes the solution.

6. Retention, encryption and data in cloud services

The strategy must establish how long copies are kept. The answer depends on the needs of the business, the obligations of your sector and internal document rules, not on a single value valid for everyone. Write a rule per data type, with a minimum and a maximum period, and have a legal adviser check the legal side. Keeping too little prevents you from recovering an error discovered late; keeping too much increases costs and the surface of risk.

Encryption must also be decided in advance. Copies should be encrypted, and the key must be kept in a separate, documented location, because an encrypted backup without its key is useless. Finally, remember that data held in cloud email and document services is not always covered by the backup of the service provider. Ask in writing which recovery functions exist, for how long they remain available and who can authorise them.

7. Verifying restores and responsibilities

A strategy does not exist until a restore has been carried out successfully. Plan periodic tests: restore of a file, a folder, a database and, at regular intervals, of a complete system in an isolated environment. Record the date, the time taken, the outcome and any problems encountered. The time measured during the test is your real RTO, and it is often different from the one stated in the supplier's documents.

Every part of the strategy needs an owner and a substitute. Who checks the outcome of backups every day? Who decides whether a restore is acceptable? Who calls the supplier out of hours? Who communicates with management during an incident? If these questions have no written answer, the strategy depends on a single person, and people change roles or leave the business. Shared responsibility without names often ends up with no one carrying it out.

8. What to read next: the topics for further study

This guide is the map of the cluster dedicated to backup design. The questions that the following guides explore are listed below. You can tackle them in a different order, but it is advisable to start with defining RPO and RTO, because all other choices depend on those objectives. Once the objectives are clear, move on to the comparison of tools and to the retention policy.

Each topic answers one part of the strategy. Read them in this order if you are starting from scratch, or choose the one that matches the problem you need to solve now. Each guide stands on its own, but the first two lay the groundwork for the others: without clear objectives and without a choice of separate copies, even the most refined technical decisions remain in suspense.

  • How to define RPO and RTO application by application, with a repeatable method
  • Differences between backup, replication and snapshots and when to use each
  • How long to keep backups and how to build a retention policy
  • Agent-based and agentless backup for virtual machines and databases
  • Backup of data in SaaS services and why the cloud suite is not enough on its own
  • Backup encryption and the risks of losing the keys
  • The role of tape today and the cases where it still makes sense
  • Sizing storage with deduplication and data growth
  • Daily checklist for operational verification

9. How to start: inventory and the first question to your supplier

The first concrete step is the inventory of systems holding valuable data, even as a simple table with name, owner and type of data. It takes a few hours and costs nothing. Then ask the people who use each system how much work they can afford to lose and how long they can stay down. Do not look for a perfect answer: a reasoned estimate is enough to start and can be improved through testing. Also note who gave each answer. Set a date for the first answers too.

Take the inventory to your IT supplier and ask three questions: which copies exist today for each system, when the last restore was tested and who is responsible for each part. The answers will show where you really stand, often differently from how you imagine it. The strategy comes from this comparison, not from a document written in isolation and then filed away. Bring this question to the next management meeting too.

Frequently Asked Questions

How long does it take to design a backup strategy?

It depends on the number of systems, the complexity of the infrastructure and how clear the internal objectives are. In a business with few systems the inventory and the objectives can be defined in a few working sessions. In a more complex infrastructure the process takes longer and involves the heads of departments. There is no standard duration. Completing the decisions matters more than keeping to a fixed timetable.

Do I need an external consultant or is the IT supplier enough?

The IT supplier can propose the technical architecture, but the business objectives and the data retention obligations need input from those who know the business and, where needed, from a data protection consultant. Ask the supplier to document its technical assumptions in writing and have a professional validate the legal side. This way each party is accountable for its own part, and no decision remains implicit.

Is the 3-2-1 rule mandatory?

No. It is a widespread good practice and a useful starting point, not a legal obligation. GDPR does not prescribe a specific backup technique, but it requires measures appropriate to the risk, including the ability to restore data in a timely manner. The rule helps demonstrate that you have copies independent of one another. Assess whether your architecture follows it and, if it does not, record the reason and the decision taken.

Is cloud backup enough on its own?

It depends on how it is built. A cloud copy separated from production systems can be part of a good strategy, but it must be verified like any other copy: who can delete it, how long the restore takes and in what format it is kept. Do not assume a copy is safe just because it is remote.

How many copies should I keep?

There is no single number. It depends on the acceptable data loss, how often the data changes and retention needs. The strategy must state which copies exist, for how long they remain available and where they are held. The number comes after these choices, and the dedicated guide on retention explores the topic further.

How often should the strategy be reviewed?

It should be reviewed when systems change, when a critical service is added, when retention obligations change and after an incident or a restore test that revealed a problem. Absent such changes, an annual review is a reasonable practice, to be agreed with the business continuity manager.

About this content

General informational guide. Regulations and technical requirements change over time: always check the official sources and assess your specific case with a qualified technical or legal advisor.

GET IN TOUCH

Talk to our team.

We'll show how Altovar integrates into your infrastructure. No commitment, no sales pitch. Just a technical conversation.

BOOK A DEMO →