KMS Management
This page describes the management of a Key Management System (KMS) within the Weka system.
Overview
When creating an encrypted filesystem, a KMS must be used to properly secure the encryption keys.
The Weka system uses the KMS to encrypt filesystem keys. When the Weka system comes up, it uses the KMS to decrypt the filesystem keys and use its in-memory capabilities for data encrypting/decrypting operations.
When a snapshot is taken using the Snap-To-Object feature, the encrypted filesystem key is saved along with the encrypted data. In the event of rehydrating this snapshot to a different filesystem (or when recovering from a disaster to the same filesystem in the Weka cluster), the KMS is used to decrypt the filesystem key. Consequently, the same KMS data must be present when performing such operations.
For increased security, the Weka system does not save any information that can reconstruct the KMS encryption keys, which is performed by the KMS configuration alone. Therefore, the following should be considered:
If the KMS configuration is lost, the encrypted data may also be lost. Therefore, a proper DR strategy should be set when deploying the KMS in a production environment.
The KMS should be available when the Weka system comes up when a new filesystem is created, and from time to time when key rotations must be performed. Therefore, it is recommended that the KMS be highly available.
For more information, refer to KMS Best Practices.
The Weka system supports the following KMS types:
KMIP compliant KMS (protocol version 1.2 and up).
HashiCorp Vault version 1.1.5 up to 1.9.x (not limited to the KMIP compliant version). For setting up Vault to work with the Weka system, refer to Setting Up Vault Configuration.
Deploy one of the supported KMS types that best suit your requirements. For additional information on KMS support, contact the Weka Sales or Support Teams.
Managing KMS Using the GUI
Adding a KMS
To add a KMS to the Weka system, go to the KMS Configuration screen on the left sidebar and click Configure KMS.
The Configure KMS dialog box will be displayed.
Enter the URL, key name, and API token, and click Update to configure the KMS.
Viewing the KMS
To view the configured KMS, go to the main KMS configuration screen.
Updating the KMS Configuration
To update the KMS configuration, click Update KMS. The Configure KMS dialog box will be displayed.
Update the URL, master key or API token, and click Update.
Removing the KMS
To remove a KMS configuration (an operation that is only possible if no encrypted filesystems exist), click the Reset KMS button on the main KMS Configuration screen. The KMS Reset dialog box will be displayed.
Click Yes to remove the KMS configuration.
Managing KMS using the CLI
Adding/Updating a KMS
Command: weka security kms set
Use the following command line to add or update the Vault KMS configuration in the Weka system:
weka security kms set <type> <address> <key-identifier> [--token token] [--client-cert client-cert] [--client-key client-key] [--ca-cert ca-cert]
Parameters in Command Line
Name
Type
Value
Limitations
Mandatory
Default
type
String
Type of the KMS
Either vault
or kmip
Yes
address
String
KMS server address
URL
for Vault, hostname:port
for KMIP
Yes
key-identifier
String
Key to be used for encryption-as-a-service in the KMS
Key name (for Vault) or a key UID (for KMIP)
Yes
token
String
API token to access Vault KMS
Must have:
read permissions to
transit/keys/<master-key-name>
write permissions to
transit/encrypt/<master-key-name>
andtransit/decrypt/<masterkeyname>
permissions to
/transit/rewrap
andauth/token/lookup
Must be supplied for vault
and must not be supplied for kmip
client-cert
String
Path to the client certificate PEM file
Must permit encrypt
and decrypt
permissions
Must be supplied for kmip
and must not be supplied for vault
client-key
String
Path to the client key PEM file
Must be supplied for kmip
and must not be supplied for vault
ca-cert
String
Path to the CA certificate PEM file
Optional for kmip
and must not be supplied for vault
Note: For the add/update command to succeed, the KMS should be preconfigured and available with the key and a valid token.
For Example:
Setting the Weka system with a Vault KMS:
weka security kms set vault https://vault-dns:8200 weka-key --token s.nRucA9Gtb3yNVmLUK221234
Setting the Weka system with a KMIP complaint KMS (e.g., SmartKey):
weka security kms set kmip amer.smartkey.io:5696 b2f81234-c0f6-4d63-b5b3-84a82e231234 --client-cert smartkey_cert.pem --client-key smartkey_key.pem
Viewing the KMS
Command: weka security kms
Use this command to show the details of the configured KMS.
Removing the KMS
Command: weka security kms unset
Use this command to remove the KMS from the Weka system. It is only possible to remove a KMS configuration if no encrypted filesystems exist.
Note: To force remove a KMS even if encrypted filesystems exist, use the --allow-downgrade
attribute. In such cases, the encrypted filesystem keys are re-encrypted with local encryption and may be compromised.
Re-wrapping Filesystem Keys
Command: weka security kms rewrap
If the KMS key is compromised or requires rotation, the KMS admin can rotate the key in the KMS. In such cases, this command is used to re-encrypt the encrypted filesystem keys with the new KMS master key.
weka security kms rewrap [--new-key-uid new-key-uid]
Parameters in Command Line
Name
Type
Value
Limitations
Mandatory
Default
new-key-uid
String
Unique identifier for the new key to be used to wrap filesystem keys
Must be supplied for kmip
and must not be supplied for vault
Note: Existing filesystem keys that are part of the Snap-To-Object feature will not be automatically re-encrypted with the new KMS key.
Note: Unlike in Vault KMS, re-wrapping a KMIP-based KMS requires generating a new key in the KMS, rather than rotating the same key. Hence, the old key should be preserved in the KMS in order to be able to decrypt old Snap2Obj snapshots.
KMS Best Practices
The KMS is the only source holding the key to decrypt Weka system filesystem keys. For non-disruptive operation, it is highly recommended to follow these guidelines:
Set up DR for the KMS (backup/replication) to avoid any chance of data loss.
Ensure that the KMS is highly available (note that the KMS is represented by a single URL in the Weka system).
Provide access to the KMS from the Weka system backend hosts.
Verify the methods used by the KMS being implemented (each KMS has different methods for securing/unsealing keys and for reconstructing lost keys, e.g., Vault unsealing methods, which enable the configuration of auto unsealing using a trusted service).
Refer to Production Hardening for additional best practices suggested by HashiCorp when using Vault.
Note: Taking a Snap-To-Object ensures that the (encrypted) filesystems keys are backed up to the object store, which is important if a total corruption of the Weka system configuration occurs.
Setting-Up Vault Configuration
Enabling 'Transit' Secret Engine in Vault
As described above, the Weka system uses encryption-as-a-service capabilities of the KMS to encrypt/decrypt the filesystem keys. This requires the configuration of Vault with the transit
secret engine.
For more information, refer to Vault transit secret-engine documentation.
Setting-Up a Master Key for the Weka System
Once the transit
secret engine is set up, a master key for use with the Weka system must be created.
Note: It is possible to either create a different key for each Weka cluster or to share the key between different Weka clusters.
For more information, refer to Vault transit secret-engine documentation.
Creating a Policy for Master Key Permissions
Create a
weka_policy.hcl
file with the following content:
This limits the capabilities so there is no permission to destroy the key, using this policy. This protection is important when creating an API token.
Create the policy using the following command:
Obtaining an API Token from Vault
Authentication from the Weka system to Vault relies on an API token. Since the Weka system must always be able to communicate with the KMS, a periodic service token must be used.
Verify that the
token
authentication method in Vault is enabled. This can be performed using the following command:
To enable the token authentication method use the following command:
Log into the KMS system using any of the identity methods supported by Vault. The identity should have permission to use the previously-set master key.
Create a token role for the identity using the following command:
Note: The period
is the time set for a renewal request. If no renewal is requested during this time period, the token will be revoked and a new token must be retrieved from Vault and set in the Weka system.
Generate a token for the logged-in identity using the following command:
For more information on obtaining an API token, refer to Vault Tokens documentation.
Note: The Weka system does not automatically renew the API token lease. It can be renewed using the Vault CLI/API. It is also possible to define a higher maximum token value (max_lease_ttl)
by changing the Vault Configuration file.
Obtaining a Certificate for a KMIP-based KMS
The method for obtaining a client certificate and key and setting it via the KMS is different for each KMS. The certificate itself will be generated using OpenSSL, with some UID obtained from the KMS, e.g.:
Please consult the specific KMS documentation to create a certificate and link it to the Weka cluster in the KMS with sufficient privileges (encrypt/decrypt).
For example, for SmartKey KMS, follow similar instructions as suggested here to create a client certificate and key, and assign a certificate for Weka within SmartKey.
Last updated