Transform Secrets Engine

»Challenge

Vault's Transit secrets engine provides encryption service; however, the resulting ciphertext does not preserve the original data format or length.

Think of a scenario where an organization must cryptographically protect the personally identifiable information (PII) while preserving the data format and length. For example, the database schema expects a certain character length and/or only allow alphanumeric.

The preservation of the original data format or length may be driven by compliance with certain industry standards such as HIPAA or PCI.

»Solution

Vault Enterprise 1.4 with Advanced Data Protection module introduced the Transform secrets engine which handles secure data transformation and tokenization against the provided secrets. Transformation methods encompass NIST vetted cryptographic standards such as format-preserving encryption (FPE) via FF3-1 to encode your secrets while maintaining the data format and length. In addition, it can also be pseudonymous transformations of the data through other means, such as masking.

This prevents the need for change in the existing database schema.

»Prerequisites

To perform the tasks described in this tutorial, you need to have a Vault Enterprise v1.4 or later with Advanced Data Protection module.

An interactive tutorial is also available if you do not have a Vault environment to perform the steps described in this tutorial. Click the Show Terminal button to start.

»Policy requirements

To perform all tasks demonstrated in this tutorial, your policy must include the following permissions:

# Work with transform secrets engine
path "transform/*" {
  capabilities = [ "create", "read", "update", "delete", "list" ]
}

# Enable secrets engine
path "sys/mounts/*" {
  capabilities = [ "create", "read", "update", "delete", "list" ]
}

# List enabled secrets engine
path "sys/mounts" {
  capabilities = [ "read", "list" ]
}

If you are not familiar with policies, complete the policies tutorial.

»Transform secrets engine workflow

Transform secrets engine configuration workflow:

1. Enable the transform secrets engine
2. Create a role containing the transformations that it can perform
3. Create an alphabet defining a set of characters to use for format-preserving encryption (FPE) if not using the built-in alphabets.
4. Create a template defining the rules for value matching if not using the built-in template
5. Create a transformation to specify the nature of the data manipulation

Alphabets define a set of valid input/output UTF-8 characters to be used when you perform FPE. In this step, you are going to leverage one of the built-in alphabets. Read the create custom alphabets section to learn how to define your own alphabets.

Data transformation templates are constructed of type (regex), pattern (regex expression) and allowed alphabet used in the input value. Currently, regex is the only supported type. The pattern defines the data format pattern. For example, the most credit card numbers would have a pattern that can be expressed as (\d{4})-(\d{4})-(\d{4})-(\d{4}) in regex.

In this step, the use of the builtin/creditcardnumber template is demonstrated. Read the create custom templates section to learn how to define your own templates.

Transformations define the transformation template, tweak source or the masking character to be used to transform the secrets.

»Tweak source types:

»Create a new transformation

In this section, you are going to:

1. Enable the transform secrets engine
2. Create a payments role which includes card-number transformation
3. Create a card-number transformation which performs format preserving encryption

»Transform secrets

The Vault client applications must have the following in their policy to perform data encoding and decoding using the Transform secrets engine enabled at transform/.

# To request data encoding using any of the roles
# Specify the role name in the path to narrow down the scope
path "transform/encode/*" {
   capabilities = [ "update" ]
}

# To request data decoding using any of the roles
# Specify the role name in the path to narrow down the scope
path "transform/decode/*" {
   capabilities = [ "update" ]
}

$ vault write transform/encode/payments value=1111-2222-3333-4444

Key              Value
---              -----
encoded_value    8492-9808-1939-2623

$ vault write transform/decode/payments \
        value=8492-9808-1939-2623

Key              Value
---              -----
decoded_value    1111-2222-3333-4444

»Create custom templates

Templates define the data format patterns that you wish to keep while transforming the secrets. In this section, you are going to create a transformation template which encodes British passport numbers.

British passport has a pattern of 9-digit numeric value which can be expressed as (\d{9}) using the regex. The parentheses tell Vault to encode all values grouped within; therefore, (\d{9}) will encode the entire passport number.

$ vault list transform/template

Keys
----
builtin/creditcardnumber
builtin/socialsecuritynumber

$ vault write transform/template/uk-passport-tmpl \
    type=regex \
    pattern="(\d{9})" \
    alphabet=builtin/numeric

$ vault write transform/transformations/fpe/uk-passport \
    template=uk-passport-tmpl \
    tweak_source=internal \
    allowed_roles=*

$ vault write transform/role/payments transformations=card-number,uk-passport

$ vault write transform/encode/payments value="123456789" \
    transformation=uk-passport

Key              Value
---              -----
encoded_value    128151714

»Create custom alphabets

Alphabet defines a set of characters (UTF-8) that is used for FPE to determine the validity of plaintext and ciphertext values.

These are a number of built-in alphabets available to use.

New alphabets can be created to satisfy the template requirements.

To learn the command, create a non-zero-numeric alphabet which contains non-zero numbers.

$ vault list transform/alphabet

Keys
----
builtin/alphalower
builtin/alphanumeric
builtin/alphanumericlower
builtin/alphanumericupper
builtin/alphaupper
builtin/numeric

$ vault write transform/alphabet/non-zero-numeric alphabet="123456789"

»Data masking

Data masking is used to hide sensitive data from those who do not have a clearance to view them. For example, this allows a contractor to test the database environment without having access to the actual sensitive customer information. Data masking has become increasingly important with the enforcement of General Data Protection Regulation (GDPR) introduced in 2018.

The following steps demonstrate the use of masking to obscure your customer's phone number since it is personally identifiable information (PII).

Now, create a phone-number-tmpl template which masks phone numbers with its country code visible.

$ vault write transform/template/phone-number-tmpl type=regex \
    pattern="\+\d{1,2} (\d{3})-(\d{3})-(\d{4})" \
    alphabet=builtin/numeric

$ vault write transform/transformations/masking/phone-number \
    template=phone-number-tmpl \
    masking_character=# \
    allowed_roles=*

$ vault write transform/role/payments \
    transformations=card-number,uk-passport,phone-number

$ vault write transform/encode/payments value="+1 123-345-5678" \
    transformation=phone-number
Key              Value
---              -----
encoded_value    +1 ###-###-####

»Batch input processing

When you need to encode more than one secret value, you can send multiple secrets in a request payload as batch_input instead of invoking the API endpoint multiple times to encode secrets individually.

»Example Scenario 1:

You received a credit card number, British passport number and a phone number of a customer and wish to transform all these secrets using the payments role.

Create an API request payload with multiple values, each with the desired transformation.

$ tee input-multiple.json <<EOF
{
  "batch_input": [
    {
      "value": "1111-1111-1111-1111",
      "transformation": "card-number"
    },
    {
      "value": "123456789",
      "transformation": "uk-passport"
    },
    {
      "value": "+1 123-345-5678",
      "transformation": "phone-number"
    }
  ]
}
EOF

Encode all the values with the payments role.

$ curl --header "X-Vault-Token: $VAULT_TOKEN" \
     --request POST \
     --data @input-multiple.json \
     $VAULT_ADDR/v1/transform/encode/payments | jq ".data"
{
  "batch_results": [
    {
      "encoded_value": "7998-7227-5261-3751"
    },
    {
      "encoded_value": "908547441"
    },
    {
      "encoded_value": "## ###-###-####"
    }
  ]
}

»Example Scenario 2:

An on-premise database stores corporate card numbers and your organization decided to migrate the data to another database. You wish to encode those card numbers before storing them in the new database.

Create a request payload with multiple card numbers.

$ tee payload-batch.json <<EOF
{
  "batch_input": [
    { "value": "1111-1111-1111-1111", "transformation": "card-number" },
    { "value": "2222-2222-2222-2222", "transformation": "card-number" },
    { "value": "3333-3333-3333-3333", "transformation": "card-number" },
    { "value": "4444-4444-4444-4444", "transformation": "card-number" }
  ]
}
EOF

Encode all the values with the payments role.

$ curl --header "X-Vault-Token: $VAULT_TOKEN" \
    --request POST \
    --data @payload-batch.json \
    $VAULT_ADDR/v1/transform/encode/payments | jq ".data"
{
  "batch_results": [
    {
      "encoded_value": "7998-7227-5261-3751"
    },
    {
      "encoded_value": "2026-7948-2166-0380"
    },
    {
      "encoded_value": "3979-1805-7116-8137"
    },
    {
      "encoded_value": "0196-8166-5765-0438"
    }
  ]
}

To decode these values.

First, create a request payload with the encoded card numbers.

$ tee payload-batch.json <<EOF
{
  "batch_input": [
    { "value": "7998-7227-5261-3751", "transformation": "card-number" },
    { "value": "2026-7948-2166-0380", "transformation": "card-number" },
    { "value": "3979-1805-7116-8137", "transformation": "card-number" },
    { "value": "0196-8166-5765-0438", "transformation": "card-number" }
  ]
}
EOF

Decode all the values with the payments role.

$ curl --header "X-Vault-Token: $VAULT_TOKEN" \
    --request POST \
    --data @payload-batch.json \
    $VAULT_ADDR/v1/transform/decode/payments | jq ".data"
{
  "batch_results": [
    {
      "decoded_value": "1111-1111-1111-1111"
    },
    {
      "decoded_value": "2222-2222-2222-2222"
    },
    {
      "decoded_value": "3333-3333-3333-3333"
    },
    {
      "decoded_value": "4444-4444-4444-4444"
    }
  ]
}

»Next steps

To actually integrate your application with Vault and leverage the transform secrets engine, there are a number of resources must be configured.

Before the application can even request data transformation, it first needs to authenticate with Vault. Therefore, an auth method (e.g. AWS, Kubernetes, AppRole) must be enabled and configured for the application to use. In addition, an appropriate policy must be created and attached to the client token.

You can codify the Vault configuration using Terraform, and make the configuration repeatable. Terraform Vault Provider supports transform secrets engine. It can create policies, enable and configure auth methods and more.

Refer to the Codify Management of Vault Enterprise Using Terraform tutorial to learn how to leverage Terraform.

On the application side, you can run Vault Agent to authenticate with Vault and manage the lifecycle of the client token. Refer to the following tutorials to learn more about Vault Agent:

• Vault Agent with AWS
• Vault Agent with Kubernetes

The Encrypting Data while Preserving Formatting with the Vault Enterprise Transform Secrets Engine blog post introduces some code examples to invoke the transform secrets engine using Vault API.

»Summary

The Transform secrets engine performs secure data transformation and tokenization against the input data. Transformation methods may encompass NIST vetted cryptographic standards such as format-preserving encryption (FPE) via FF3-1, but can also be pseudonymous transformations of the data through other means, such as masking. This tutorial walked through the use of the Transform secrets engine step-by-step.

»Limits:

The Transform secret engine obeys the FF3-1 minimum and maximum sizes on the length of an input, which are a function of the alphabet size.

»Help and Reference

• Transform Secrets Engine (API)
• Transform Secrets Engine
• Encrypting Data while Preserving Formatting with the Vault Enterprise Transform Secrets Engine