Creating custom shielded images

This topic describes how to prepare the disk, generate security certificates, and enable any necessary operating system (OS) features to create a custom shielded image.

By default, Shielded VM supports Container-Optimized OS, various distributions of Linux, and multiple versions of Windows Server. But if you require custom images for your application, you can still take advantage of Shielded VM.

Preparing the disk

Shielded VM relies on Unified Extensible Firmware Interface (UEFI)-compliant firmware to support features such as Secure Boot. Shielded VM requires a GUID Partition Table (GPT) scheme; master boot record (MBR) is not supported.

The disk must have at least two partitions:

  • EFI System Partition (ESP): 100 megabytes (MB) is sufficient for this partition, and is only a suggestion. You can create a larger partition if necessary. The only requirement for the ESP is that it should be formatted with a File Allocation Table (FAT) filesystem.
  • OS Partition: The rest of the disk. This partition contains the boot OS (Linux or Windows). There is no restriction on the size of this partition.

You can create more data partitions as necessary.

Copying the OS to the OS partition

Once the disk is formatted and partitioned correctly, copy the OS files into the OS partition. The OS has a boot loader that must be located at a valid path on the ESP, as specified in the UEFI specification: \EFI\Boot\bootx64.efi. Note that it may be necessary to copy the OS boot loader to the given location.

For Windows, there is command called bcdboot that can be used to copy the OS boot loader to the correct location, in addition to other actions that Windows requires (such as copying the BCD store). For more information, see BCDBoot Command-Line Options on the Microsoft Hardware Dev Center.

When using Shielded VM images, you can also take advantage of two additional security features: virtual Trusted Platform Module (vTPM) and integrity monitoring. The following sections outline the benefits of these features and the OS requirements.

Virtual Trusted Platform Module (vTPM)

A trusted platform module is a specialized device to protect objects, like keys and certificates, that you use to authenticate access to your system. On Shielded VM images, virtualized versions of TPM devices are used to enable Measured Boot. In brief, Measured Boot ensures the integrity of the critical load path of boot and kernel drivers. vTPM and Measured Boot are covered in more detail in the Shielded VM documentation.

In order to take advantage of the vTPM and Measured Boot, a driver is required. The minimum OS versions with TPM 2.0 support are:

  • Windows Server 2012
  • Linux version 3.20
  • Red Hat Enterprise Linux 7.3

Integrity monitoring

Integrity monitoring provides a way to understand and make decisions about the state of your VM instances. Monitoring uses the data generated by Measured Boot to report on the VM instance. The Shielded VM documentation has more information about integrity monitoring and automating responses to integrity validation failures.

To support the Shielded VM integrity monitoring feature, the image must produce integrity signals:

  • Windows generates integrity signals by default.
  • Linux must have the Integrity Measurement Architecture (IMA) module installed and enabled. The module must have CONFIG_IMA_MEASURE_PCR_IDX set to 10. This is the default value for the IMA module.

Importing the disk image to Compute Engine

Once the image is prepared, you have to upload the image into Compute Engine. For the necessary steps to upload the image to Google Cloud, see Importing boot disk images to Compute Engine.

Setting up certificates for Secure Boot

When adding a Shielded VM image, a set of Secure Boot public certificates and databases are passed into Compute Engine. These files are stored in the corresponding UEFI variables and used to establish trust relationships between the platform, firmware, and OS. Certificates are Distinguished Encoding Rules (DER)-encoded X.509 certificates. The databases can be either a certificate or raw binary. There are four values in total:

  • Platform Key (pk): A key used to establish the trust relationship between the platform owner and the firmware. You may only specify one platform key, and it must be a valid X.509 certificate.
  • Key Exchange Key (kek): A key used to establish a trust relationship between the firmware and the OS. You may specify multiple keys for this value.
  • Forbidden Key Database (dbx): A database of certificates that have been revoked and will cause the system to stop booting if a boot file is signed with one of them. You may specify single or multiple values for this value.
  • Key Database (db): A database of certificates that are trusted and can be used to sign boot files. You may specify single or multiple values for this value.

The UEFI specification contains more information about these values and how they work.

In the following example OpenSSL is used to create the Secure Boot keys and certificates.

  • Generate a 2048-bit RSA key pair

      $ openssl genrsa -out secure-boot-key.rsa 2048
  • Generate a self-signed X.509 certificate from the key in DER format

      $ openssl req -new -x509 -sha256 -subj '/CN=secure-boot' -key secure-boot-key.rsa
      -outform DER -out secure-boot-cert.pem

Adding the shielded image to Google Cloud

Using the uploaded image and certificates, you can now add the image to Compute Engine. The image can be added using the Google Cloud CLI or the Compute Engine API.


Add the custom image to Compute Engine:

gcloud compute images create [IMAGE_NAME] \
--source-disk [SOURCE_DISK] \
--source-disk-zone [ZONE] \
--platform-key-file=<file.der> \
--key-exchange-key-file=<file.der> \
--signature-database-file=<file.bin>,<file.der> \
--forbidden-database-file=<file.bin> \


  • [IMAGE_NAME] is the name for the new image.
  • [SOURCE_DISK] is the disk from which you want to create the new image.
  • [ZONE] is the zone where the disk is located.

The WINDOWS option for guest-os-features is only required when using a Windows image. For more information on creating an image, see the gcloud create reference.


Follow the instructions to create an image from a persistent disk but specify the initial_state_config in the request body.

"sourceDisk": "/zones/[ZONE]/disks/[SOURCE_DISK]",

"initial_state_config": {
    "pk": {
        "content": [KEY],
        "fileType": [BIN,X509]
    "keks": [
            "content": [KEY],
            "fileType": [BIN,X509]
    "dbxs": [
            "content": [KEY],
            "fileType": [BIN,X509]
    "dbs": [
            "content": [KEY],
            "fileType": [BIN,X509]

Default certificates

Note that pk, keks, dbxs and dbs are optional fields. If you provide an initial state configuration, some or all of these fields may be unset. When a new instance is created from the image, Google Cloud provides a default value for PK, KEK, db, and dbx unless a custom value was set on any unset field. If you provide no initial state configuration (that is, the configuration is missing, not just empty), the image will have the initial state configuration of the source image.

These fields' default values are:

  • PK: The certificate associated with the default private key created by Google.
  • KEK: The default Microsoft KEK certificate. Download from Microsoft: MicCorKEKCA2011_2011-06-24.crt.
  • dbx: The default Microsoft DBX revocation list. Download from Unified Extensible Firmware Interface Forum: UEFI revocation list file
  • db: The following two certificates:
    • The Microsoft Windows Production PCA 2011 with a SHA-1 Cert Hash of 58 0a 6f 4c c4 e4 b6 69 b9 eb dc 1b 2b 3e 08 7b 80 d0 67 8d. Download from Microsoft: MicWinProPCA2011_2011-10-19.crt.
    • The Microsoft Corporation UEFI CA 2011 with a SHA-1 Certificate Hash of 46 de f6 3b 5c e6 1c f8 ba 0d e2 e6 63 9c 10 19 d0 ed 14 f3. Download from Microsoft: MicCorUEFCA2011_2011-06-27.crt.

Be careful, since adding your own certificates will overwrite the default ones rather than merging them with the ones you provide.