Many organisations enable backup encryption with a single click, tick the box in the console and consider the matter closed. The problem surfaces months later, at the worst possible moment: during an emergency restore, when they discover that nobody knows where the password used to encrypt that copy is. At that instant, the backup stops being an insurance policy and turns into a perfectly useless file.
Backup encryption is not a checkbox; it is a process that has as much to do with key management as with the cryptographic algorithm. Strong encryption with a poorly guarded key is as dangerous as no encryption at all — the risk simply changes shape: you no longer worry about a third party reading your data, you worry about not even being able to recover it yourself.
In this article we focus specifically on encrypting backups and on the discipline that underpins it: where it makes sense to encrypt, how it interacts with deduplication and immutability, how Veeam implements it, and above all which key-management practices prevent the disaster of losing access to your own data.
Why Encrypt Backups
A backup is, by definition, a complete, consolidated copy of your most valuable data: databases, mailboxes, file systems, configurations. It is the most attractive target there is. If an attacker steals a tape, walks off with a disk or compromises the backup repository, they do not get a fragment of current information — they get your entire historical dataset in one go. An unencrypted backup repository is a massive data breach waiting to happen.
The risk multiplies the moment copies leave your premises. Encryption is practically mandatory for any offsite or cloud backup, where data travels over networks you do not control and rests on a third party's infrastructure. The regulatory framework demands it too: the GDPR names encryption as one of the appropriate technical measures to protect personal data, and Spain's National Security Framework (ENS) requires it for sensitive information both in transit and at rest. An incident affecting encrypted data usually falls outside the breach notification obligation, precisely because the data remains unreadable.
Encrypting backups does not conflict with other security layers and does not replace protecting data in production; it is an indispensable complement. You can dig into the broader context in our article on data encryption and privacy, but what concerns us here is the specific case of backup copies.
Where to Encrypt: Source, At Rest and In Transit
Not all encryption is equal, and the point where it is applied completely changes the trust model. There are three moments worth distinguishing:
Source-side or client-side encryption is applied in your own environment, before data leaves your network. The copy is encrypted on the backup server or agent, and only then travels to the destination. It is the ideal model for offsite and cloud copies because it means the provider never sees the data in the clear: even if their storage were compromised, all they would find is encrypted blocks they cannot decrypt. The trade-off is that responsibility for the key rests entirely with you.
At-rest encryption at the destination is applied on the repository itself or on the storage array (for example, disk-level or file-system encryption). It protects against physical theft of a disk or hardware decommissioning, but the data arrives at the destination in the clear and it is typically the provider or the storage system that manages the key. It is a useful layer, but it does not deliver the privacy that client-side encryption offers.
Encryption in transit (via TLS) protects data while it is being transmitted over the network, preventing anyone from intercepting it along the way. It is complementary, not alternative: you need TLS and at-rest encryption. One protects the journey; the other, the store. In every case, the de facto standard is AES-256, a robust, widely audited symmetric algorithm that, with a well-managed key, is infeasible to break by brute force with today's technology.
Table: Where to Encrypt, What It Protects Against and Key Implications
The following table summarises the three encryption layers, the specific threat each one mitigates and who carries responsibility for the key:
| Where it is encrypted | Protects against | Key consideration |
|---|---|---|
| At source / client | Provider or third party reading the data; theft of the remote repository | The key is yours: if you lose it, nobody can help you |
| At rest (destination) | Physical theft of disks or tapes; hardware decommissioning | Usually managed by the provider or array; less privacy from the provider |
| In transit (TLS) | Interception of the copy while it travels over the network | Ephemeral session keys managed by the protocol; does not protect at rest |
The practical conclusion is clear: for offsite backup, client-side encryption offers the maximum confidentiality protection, but it shifts all responsibility for the key onto your organisation. And that is where the real problems begin.
The Encrypted Backup Paradox: Key Management
Here is the crux of the matter. The very mechanism that makes an encrypted backup safe from third parties is the one that renders it useless if you lose the key. AES-256 has no backdoor, no master key, no recovery path: it is mathematically infeasible to decrypt the data without the correct key. That is exactly the property you want to protect yourself from an attacker, and also the one that dooms you if you are the one who loses access.
Critical warning:
If you lose the encryption key or passphrase, your backup is unrecoverable garbage. Neither you, nor your provider, nor the software vendor, nor any recovery tool can decrypt it. There is no "forgot your password?" for AES-256. Countless companies have discovered their copies were inaccessible at the exact moment they needed them to recover from an incident.
The scenario repeats itself with depressingly similar variations: the person who set up encryption left the company and the password died with them; the key was stored in a document inside the very server that got encrypted by ransomware; or the passphrase was jotted on a sticky note that nobody can find after the office move. In every case, the outcome is identical: petabytes of corporate history turned into random noise.
That is why the fundamental maxim is counter-intuitive: in an encrypted backup, protecting the key is as important as protecting the data. Key availability is part of your RPO and RTO just as much as the copy itself. A recovery plan that does not account for how the encryption key is recovered simply is not a recovery plan.
Key Management Best Practices
Managing backup encryption keys well comes down to a set of principles that the NIST key management project formalises and that are worth applying with discipline:
- check_circle Separate the key from the data: the key or passphrase must never be stored inside the system that encrypts, and much less inside the backup itself. If the compromised server also holds the key, encryption has protected you from nothing.
- check_circle Secure, centralised custody: use a secrets manager, an HSM (Hardware Security Module) or a KMS (Key Management Service) to store and control access to keys. For master recovery keys, a sealed envelope in a physical safe is still a valid last-resort option.
- check_circle Key rotation: change encryption keys periodically to limit the exposure window if a key is ever compromised, always retaining the previous keys needed to restore older copies.
- check_circle Escrow and controlled recovery: implement a key escrow mechanism that allows the key to be recovered in an audited way in an emergency, without depending on one person's memory.
- check_circle Do not depend on a single person: apply the principle of dual control or split knowledge. If only the administrator who left knew the password, you have a single human point of failure at the heart of your recovery plan.
If your organisation is going to handle keys at scale, it is worth adopting a dedicated secrets store rather than improvising with spreadsheets. We cover how to approach it in our guide to secrets management with Vault.
Encryption, Deduplication, Immutability and Air Gap
Encryption does not live in isolation within your backup strategy; it interacts with other functions, and you need to understand those interactions to avoid surprises.
The first is deduplication. Encrypted data looks random, and two identical blocks encrypted with different nonces produce different results. That is why encrypting before deduplicating destroys the savings ratios: the deduplication engine can no longer find repeated blocks to eliminate. The correct order is to deduplicate and compress first, then encrypt. We develop this in detail in the article on deduplication and compression in backup.
The second is the relationship with immutability and air gap. It is a common misconception to think encryption protects against ransomware. It does not: encryption protects confidentiality (nobody can read the data), but it does not stop a copy from being deleted or overwritten. Against an attack that wants to destroy your backups, what you need is immutability (copies that cannot be modified or deleted for a period) and air gap (isolating the copy from the network). Encryption and immutability are complementary: the first protects the secret, the second protects the existence of the data. A solid strategy needs both.
How Veeam Handles It
Veeam implements encryption at the backup job level with AES-256. When you enable it, you define an encryption password from which the key that protects that job's data is derived. Data is encrypted on the source side before being written to the repository, which makes it the ideal model for copies heading to an external destination.
For environments that want to avoid the single point of a manual password, Veeam allows a KMS to be integrated via KMIP, delegating key generation and custody to a specialised service. In either model, Veeam's warning is explicit and matches everything above: store the encryption password in a safe, separate place. If it is lost and there is no configuration server from which to recover the key, restoring those copies becomes impossible.
Veeam offers a password recovery mechanism tied to the configuration backup of Veeam Backup & Replication itself, but that only shifts the problem: now you have to protect and be able to restore that configuration. The rule does not change: the key, or the means to recover it, must exist outside the system being protected.
EasyDataHost: Offsite Backup with Client-Side Encryption
At EasyDataHost we run offsite backup with Veeam Cloud Connect from our own datacenter in Spain, with ISO 27001 certification and ENS compliance. Our model relies on client-side encryption:
- arrow_right Copies are encrypted in your environment, before they leave your network, so they reach our repository already encrypted with your key.
- arrow_right We do not access your data in the clear: as the provider, we only hold encrypted blocks we cannot decrypt, guaranteeing confidentiality against any access to the storage.
- arrow_right We combine encryption in transit (TLS) with immutable copies at the destination, adding confidentiality and protection against deletion within a single strategy.
- arrow_right We support you in designing key management: where to store keys, how to document recovery and how to avoid the single human point of failure.
This model gives you the best of both worlds: the privacy of client-side encryption and the peace of mind of managed infrastructure. That said, responsibility for safeguarding the key remains yours, which is why we insist so much on getting it right from day one.
Frequently Asked Questions
What happens if I lose my backup encryption key?
If you lose the key or passphrase, the encrypted backup is unrecoverable: neither you, nor the provider, nor any tool can decrypt it, because AES-256 has no backdoor. That is why the key must be stored outside the system that encrypts and outside the backup itself, with secure custody and at least one controlled backup copy.
Where should I encrypt the backup, at source or at destination?
For offsite or cloud copies the ideal is to encrypt at source, on the client, before data leaves your network, so the provider never sees the data in the clear. At-rest encryption at the destination protects the repository against physical theft, but in that model it is usually the provider who holds the key.
Does encryption protect against ransomware that deletes backups?
Not on its own. Encryption protects confidentiality, meaning nobody can read the data, but it does not stop a copy from being deleted or overwritten. For that you need immutability and air gap, which are complementary measures to encryption, not substitutes.
Conclusion
Encrypting backups is essential, but the algorithm is the trivial part: what separates a solid strategy from a latent disaster is key management. These are the ideas worth taking away:
- arrow_right Always encrypt offsite and cloud copies with AES-256, preferably client-side, so the provider never sees your data in the clear and to comply with GDPR and ENS.
- arrow_right The key is as critical as the data: without it, an encrypted backup is unrecoverable. Key availability is part of your recovery plan.
- arrow_right Keep keys outside the encrypted system, with a KMS/HSM or secrets manager, rotation, escrow and no dependence on a single person.
- arrow_right Combine encryption with immutability and air gap: encryption protects confidentiality; immutability protects against deletion. They are complementary, not interchangeable.
If you want to review how you encrypt and safeguard the keys of your copies, or design an offsite backup with client-side encryption and immutability, contact our team for a technical assessment with no obligation.