GUIDE · BACKUP DESIGN

Sizing backup storage: deduplication, data growth and backup windows

After this guide you will know how to estimate the capacity your backup storage needs, starting from real data and taking retention, growth, deduplication and backup windows into account. You will see how to build a reasonable margin and how to avoid the most common underestimates, which lead to interrupted backups precisely when they are needed.

Updated october 2026
7 min read
Altovar Technical Editorial Team

1. Why backup sizing almost always goes wrong

Backup storage is often sized on the basis of current data, without considering how much it will grow or how many copies must remain available. The result is a system that works well for a few months and then fills up, with copies failing or having to be deleted in a hurry to make room. Sizing should be thought of as an estimate over several years, not a snapshot of the moment. Even a small initial error, repeated over time, produces a very different result. For this reason it is worth reviewing the calculation at regular intervals, not only when space runs out.

The most frequent mistakes are three: forgetting the multiple copies required by the retention policy, not accounting for data growth and ignoring the space taken up by the temporary data of jobs. Each of these can lead to a significant underestimate. A simple, documented method reduces the risk and makes the choices verifiable. The method should be written down once and reused at every review, so that comparisons over time remain consistent.

2. The data to collect

Before any calculation you need real data. For each system, collect the current size of the data to be protected, the rate at which it changes and the amount of data that changes each day. Also ask what retention is planned for each category, because it is the factor that multiplies the space required. Without these elements, any figure is just an assumption. Ask the owner of each system for the estimate they have in mind, and compare it with the measured data.

Also collect information on the composition of the data: how many files are already compressed, how many are images or videos, how many machines share an operating system or a similar application. These elements determine how much deduplication and compression can reduce the space, and they should be measured, not assumed. Also note the systems that change in an irregular way, because they are the ones that most often surprise. Ask too when the last major changes were made.

3. The elements of the calculation

The calculation starts from the data to be protected and multiplies it by the copies that the retention policy requires, taking growth into account. With deduplication, meaning the removal of identical data blocks, and compression, the effective space can drop significantly, but the reduction depends on the data and is not guaranteed. For this reason the calculation should be made with a prudent margin and verified with a test on a real sample.

The table summarises the components to measure, with the most common error for each. Use it as a checklist when preparing the calculation. Every row should be completed with real data, not generic estimates, otherwise the result loses its value. If a row stays empty, the calculation is incomplete and the margin is unreliable. Then compare the total with the space currently available.

ComponentWhat it measuresCommon error
Source dataCurrent volume of data to protect, per systemUsing the volume from a single period or forgetting some systems
GrowthHow much data increases over the planning periodCalculating only on the current situation
RetentionHow many historical copies remain availableIgnoring periodic full copies and multiple versions
Deduplication and compressionHow much the space shrinks on real dataApplying theoretical reductions from other environments
Temporary dataSpace used during jobs and maintenance operationsNot allowing working space for the largest jobs

4. The backup window and bandwidth

Space is not the only constraint. The copy must complete within the planned window, that is, in the period when the load on the systems and the network is acceptable. If the data to be copied grows faster than the available bandwidth, the window stretches until it overlaps with working hours. Sizing must therefore include write speed and network capacity, not just the volume to be stored.

Check the window for the largest system and for peak days, such as the close of an accounting period or a major update. A copy that completes on time under normal conditions may exceed the window when data grows. Plan a time margin and, if needed, split the jobs or move the heaviest copies to dedicated time slots.

5. Margins, growth and review

A reasonable margin is needed to absorb growth, temporary peaks and unexpected events. The value of the margin is not universal: it depends on how predictable the data is and how easily capacity can be increased. A stable environment can accept a tighter margin, while one with strong growth needs a wider one. Write down the reasoning behind the chosen margin, so that whoever reviews the plan understands the logic.

Sizing should be reviewed periodically, comparing actual growth with the forecast. When the deviation is significant, the plan should be updated before the space becomes critical. A storage threshold alert, configured with the supplier, is the simplest way to avoid being caught unprepared. Set at least two thresholds, one for attention and one for action. Whoever receives the alert must know what to do.

6. Steps to size the storage

The process is sequential and must be repeated every time the data or the retention rules change. Start from the inventory and the retention policies, then measure the data and its growth, and finally calculate the capacity with a documented margin. The result should be verified with a test on a real sample before buying or allocating new resources.

  1. List the systems to be protected with the current data volume.
  2. Collect the growth rate observed over the most recent periods available.
  3. Derive from the retention policy the number and type of copies to keep.
  4. Estimate the effect of deduplication and compression on a real sample.
  5. Add the working space for jobs and maintenance operations.
  6. Apply a documented margin and check the copy window on peak days.

7. How to start: measure before buying

This week, measure the current data volume of the most important systems and the space already used by the backup storage. Ask the supplier to show you the growth over recent periods and the configured retention. With these three figures you already have a first estimate and understand how far you are from a problem. Also note the date of the measurement, so you can compare it with later ones.

Then prepare a simple spreadsheet with the components from the table, to be updated every quarter. Perfection is not needed at the first attempt: what matters is a repeatable method that tells you when to act. Before making a purchase or an allocation, verify the result with a test on a real sample. If the result changes a lot between two measurements, look for the cause before deciding.

Frequently Asked Questions

How is data growth estimated?

By looking at the actual trend over time, not a hypothetical figure. Measure the volume of protected data at regular intervals and compare the curve with ongoing projects and planned activities. If the business is growing, that growth should be included in the plan. If the data is stable, a smaller margin is sufficient. Document your assumptions so that they can be corrected.

Does deduplication solve the space problem?

Deduplication helps a great deal in some cases, such as similar machines or repeated copies of data that changes little, but it is not a guarantee. On data that is already compressed or very different from one item to another, the reduction may be modest. Do not base sizing on theoretical reductions: measure them on a real sample and treat deduplication as an advantage to be verified, not a certainty.

How much margin should be left free?

There is no universal percentage. The margin depends on how predictably the data grows, how quickly capacity can be increased and how critical the backups are. Write down the choice, the reason and the alert threshold, and review it when the situation changes. A reasonable margin avoids emergencies, but an excessive one leads to waste.

Does cloud storage eliminate the capacity problem?

It reduces the physical space problem, but not the issues of cost, bandwidth and restore time. A remote archive grows with the data and the retention like any other storage, and bandwidth affects both the copy window and the speed of return. Sizing remains necessary even with remote solutions, using the same quantities to measure.

How often should the calculation be redone?

Every quarter is a reasonable cadence for many organisations, with an extraordinary review when systems, retention or volumes change significantly. A storage threshold alert helps you act before the problem becomes critical. The calculation should always be redone after a major expansion or the introduction of a new service with large amounts of data.

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 →