In an audit, the sentence “our laptops are encrypted” doesn’t end the first question, it starts it: Which data is protected, with which algorithm and which key management, and how do you measure that this applies on all devices? Almost every compliance framework asks for encryption; and it expects the evidence from you. Below we explain the technical implementation of endpoint, server, database and network traffic encryption, and the verification method for each.
What compliance frameworks expect from encryption
ISO/IEC 27001:2022 Annex A 8.24 (Use of cryptography) covers defining rules for cryptography and key management. NIS2 lists cryptography and encryption policies as a separate area among risk management measures; how this area relates to the other measures is mapped in the NIS2 technical measures guide. The VDA ISA catalog that TISAX is based on questions the protection of devices carrying information and of information transferred over the network; for information with high protection needs, the expectation rises. The common denominator: the frameworks don’t require a specific product. They require a defined standard, complete coverage and measurable implementation.
That’s why the first job is to write an encryption standard that maps data types to controls:
| Data / environment | Check | Verification |
|---|---|---|
| Laptop and desktop disks | BitLocker / FileVault, with TPM | MDM compliance report |
| Server disks, virtual machines | BitLocker, LUKS, hypervisor VM encryption | Per-server status output |
| Database files and backups | TDE or the engine’s at-rest encryption | Encryption status query |
| Removable media | BitLocker To Go or a ban on use | Device control policy |
| Web and API traffic | TLS 1.2 minimum, TLS 1.3 preferred | TLS scan output |
| Management protocols | SSH, HTTPS, SNMPv3 authPriv | Configuration audit |
Data at rest: disk, database, file level
Disk-level encryption protects data when a device is lost or a disk is removed and someone tries to read it on another system. It provides no protection against an attacker running code while the system is on and a user session is active. Write this limit into the risk assessment correctly; the auditor will ask. For BitLocker on Windows, the minimum standard should include: TPM-based protection, the encryption method defined via Group Policy or Intune (for example XTS-AES 256), fixed data drives also covered, and the recovery key backed up centrally. If you use LUKS on Linux servers, plan in advance how the password prompt at boot will be handled (TPM or network-based automatic unlocking). A server in a remote data center restarting and waiting for a password at the console is a cause of outages we often run into in the field.
# Windows: BitLocker status of all volumes
Get-BitLockerVolume |
Select-Object MountPoint, VolumeStatus, EncryptionMethod, ProtectionStatus, KeyProtector
# Classic command-line equivalent
manage-bde -status
# Linux: view LUKS volumes and header information
lsblk -f
sudo cryptsetup luksDump /dev/sda3
Database encryption (for example Microsoft SQL Server TDE) protects data and log files and the backups taken from them. It targets a risk disk encryption can’t see: copying a database file or backup from a running system. Turning on TDE without backing up the certificate and private key means a backup that can’t be restored to another server during disaster recovery. The backup exists. It won’t open.
-- SQL Server: TDE status per database (3 = encrypted)
SELECT DB_NAME(database_id) AS database_name,
encryption_state,
key_algorithm, key_length
FROM sys.dm_database_encryption_keys;
When TDE is turned on for any database on the server, tempdb is also encrypted automatically and shows up in the query output; don’t interpret it in the evidence report as a separate implementation decision.
File- and application-level encryption (encryption with sensitivity labels, in-application field encryption) protects data from unauthorized people even while an authorized session is open. This is the most fine-grained control; in return, its key management and usability cost is also the highest. The risk assessment determines which data types require this level.
Data in transit: a TLS configuration standard
The TLS standard should be a written document and define at least the following:
- Protocol version: SSL 2.0/3.0, TLS 1.0 and TLS 1.1 off. TLS 1.0 and 1.1 were formally deprecated by RFC 8996. TLS 1.2 minimum; TLS 1.3 (RFC 8446) enabled on systems that support it.
- Cipher suites: For TLS 1.2, ECDHE key exchange providing forward secrecy and AEAD ciphers such as AES-GCM or ChaCha20-Poly1305. RC4, 3DES, NULL, EXPORT and anonymous suites off.
- Certificate: At least a 2048-bit RSA or ECDSA P-256 key, a SHA-256 signature, and a certificate signed by a corporate or public CA.
- Scope: Not just internet-facing web servers; internal applications, management interfaces, LDAP (LDAPS, or LDAP signing and channel binding), database connections and SMTP relay too.
The scope item is the one most often skipped. The external website gets an A grade, while the switch management interface on the internal network still speaks TLS 1.0. If TLS inspection is done on the firewall, the encryption chain between client and server is split in two; certificate and exception design is explained in the SSL/TLS inspection design guide. For options on the email side (MTA-STS, S/MIME), see the email encryption options guide.
Certificate inventory and expiry tracking
An expired certificate is both an outage and an enterprise cybersecurity risk: users get used to clicking through certificate warnings. The inventory should cover not only purchased public certificates but also those issued by the internal CA and the self-signed certificates that come by default on devices. Inventory fields: common name and SAN list, issuing CA, expiry date, key type and length, the system it’s installed on, owner and renewal method.
# Windows: machine certificates expiring within 30 days
Get-ChildItem Cert:\LocalMachine\My |
Where-Object { $_.NotAfter -lt (Get-Date).AddDays(30) } |
Select-Object Subject, NotAfter, Thumbprint
# Certificate subject and expiry date of a remote service
echo | openssl s_client -connect app.example.local:443 -servername app.example.local 2>/dev/null \
| openssl x509 -noout -subject -enddate
Under CA/Browser Forum decisions, the maximum validity of public TLS certificates is being shortened step by step; verify the current schedule with the CA/Browser Forum and your certificate provider. This trend makes manual renewal unsustainable. ACME-based automatic renewal and regular updating of the inventory through network scanning should be part of the standard.
Managing BitLocker recovery keys
If the recovery key isn’t stored centrally, encryption turns into data loss after a motherboard failure or a TPM reset. Key storage is both an availability and a security control:
- On devices joined to Entra ID, enforce backing up the key to the device object with an Intune policy; encryption shouldn’t start until the backup succeeds.
- In an Active Directory environment, keys are stored in
msFVE-RecoveryInformationobjects under the computer object; apart from administrators, give read rights only to an authorized help desk role. - Key viewing operations should be logged; a key handed to a user should be rotated afterwards (a new recovery password created).
# AD: recovery information registered for a computer (RSAT AD module)
$pc = Get-ADComputer "PC-0142"
Get-ADObject -Filter 'objectClass -eq "msFVE-RecoveryInformation"' `
-SearchBase $pc.DistinguishedName -Properties whenCreated |
Select-Object Name, whenCreated
As evidence, compare the number of encrypted devices with the number of devices whose recovery key is registered centrally. The difference is your list of devices at risk of data loss. For cases where backup data must be encrypted and its keys handled separately, the backup encryption and key management guide is a complement.
Detecting weak protocols and cipher suites
Once the standard is written, you measure whether it’s applied with network scanning. OpenSSL is enough for a quick check on a single endpoint, and Nmap’s ssl-enum-ciphers script for bulk scanning:
# Can a connection be made with TLS 1.1? (Expected to fail)
openssl s_client -connect app.example.local:443 -tls1_1 < /dev/null
# Versions and cipher suites on HTTPS and LDAPS services in the subnet
nmap -p 443,636 --script ssl-enum-ciphers 10.10.20.0/24 -oA tls-scan
# RDP security layer and encryption level (separate script)
nmap -p 3389 --script rdp-enum-encryption 10.10.20.0/24
Some current OpenSSL builds disable old protocols on the client side. So a “could not connect” result doesn’t always mean the server doesn’t support it; confirm the result with the Nmap output. On Windows servers, SCHANNEL protocol settings are managed through the registry or Group Policy; the Get-TlsCipherSuite cmdlet lists enabled cipher suites. Before turning off old protocols, identify the clients that still use them (old printers, embedded devices, old application servers) and put them on a migration plan. Otherwise the change is rolled back at the first complaint and the control stays on paper.
Proving the encryption implementation
The evidence set accepted in audits usually consists of:
- An approved encryption and key management standard (algorithms, versions, responsibilities)
- A dated disk encryption compliance report from the MDM or management tool, with an explanation of non-compliant devices
- A recovery key registration rate report
- Encryption status query output for critical databases
- The latest TLS scan report and remediation records for the weaknesses found
- The certificate inventory and evidence of the alerting mechanism for upcoming expiries
To add encryption settings to the server hardening baseline and measure them continuously, see our system and server security hardening service, and to have the current state measured independently, our control and configuration audit service.
Frequently asked questions
If BitLocker is on, is the data fully protected?
No. Disk encryption protects when a device is lost while powered off, or when a disk is removed and someone tries to read it on another system. It doesn’t protect against malware running or an unauthorized user while the system is on and a session is active; that requires access control and file-level encryption.
Is TLS 1.2 still acceptable?
Yes. Configured with secure cipher suites, TLS 1.2 is the accepted minimum version. TLS 1.0 and 1.1 were deprecated by RFC 8996. Enable TLS 1.3 on systems that support it.
Where should we store BitLocker recovery keys?
Depending on the device’s management model, on the device object in Entra ID or under the computer object in Active Directory. Encryption shouldn’t start until the key backup has succeeded; the right to read keys should be kept limited and views should be logged.
How do we prove encryption is applied on all devices?
A dated compliance report from the MDM or management tool, the recovery key registration rate, database encryption query outputs and a TLS scan report are presented together. A justified list of out-of-scope devices is also part of this set.