vSAN Encryption Services

Introduction

Data encryption is a common technique used in environments that require additional levels of security.  It consists of a process to ensure that data can only be consumed by systems that have appropriate levels of access.  Approved systems must have and use the appropriate cryptographic keys to encrypt and decrypt the data.  Systems that do not have the keys will not be able to consume the data in any meaningful way, as it will remain encrypted in accordance to the commonly used Advanced Encryption Standard (AES) from the National Institute of Standards and Technology, or NIST.

An introduction to vSAN Encryption Services

VMware vSAN offers two forms of cluster-based encryption services.

  • vSAN Data-at-Rest Encryption.  This securely encrypts all vSAN data as it lands on persistent storage devices in the hosts.  The data is non decrypted until the process of a read operation requires the data to be read.
  • vSAN Data-in-Transit Encryption.  This securely encrypts all vSAN traffic in transit across hosts.

These services can be turned on or off on a per-cluster basis, and used independently or together, and do not need or use self encrypting drives.  A list of frequently asked questions on vSAN Encryption Services can be found in the "Security" section of the vSAN FAQs.

Note that vSphere VM Encryption (sometimes referred to as "VM Encrypt") is an independent feature of vSphere that can be used on all types of storage solutions, including vSAN.  They do however share many commonalities.  The advantages and disadvantages of using VM Encrypt on vSAN will be discussed later in this document.  

 

Hypervisor Integration

vSAN Encryption services uses a native VMKernel Cryptographic module in vSphere to provide the highest levels of security and compliance.  VMware achieves FIPS 140-2 validation under the Cryptographic Module Validation Program (CMVP). The CMVP is a joint program between NIST and the Communications Security Establishment (CSE). FIPS 140-2 is a Cryptographic Modules Standards that governs security requirements in 11 areas relating to the design and implementation of a cryptographic module. vSphere and vSAN use the validated Cryptographic Module for all encryption services.

vSphere and vSAN CryptoGraphic Modules

The VMware VMkernel Cryptographic Module has successfully satisfied all requirements of these 11 areas and has gone through required algorithms and operational testing, rigorous review by CMVP and third party laboratory before being awarded certificate number 3073 by the CMVP.  Since the VMware VMkernel Cryptographic Module is part of the ESXi kernel, it can easily provide FIPS 140-2 approved cryptographic services to various VMware products and services. Virtual machines encrypted with vSphere's VM Encryption or vSAN Encryption Services work with all vSphere supported Guest Operating Systems and Virtual Hardware versions, and do not allow access to encryption keys by the Guest OS.

Implementation Considerations

Capacity Utilization

Data-at-Rest encryption in a vSphere environment can occur either inside a virtual machine such as VM Encryption, or can be accommodated by a storage system such as vSAN Data-at-Rest Encryption.  Any time that a process encrypts the data prior to space efficiency techniques like deduplication and compression occuring, this can severely impact the effectiveness of such techniques.

vSAN implements Data-at-Rest Encryption at the final step of the I/O path, which maintains its ability to use space efficiency techniques like deduplication and compression, driving down the cost of storage.  Since it is a cluster-based feature, the decision to enable will be on a per-cluster basis.  If there is a need to encrypt just a few VMs, VM Encrypt may be a fit, or perhaps using another encrypted vSAN datastore courtesy of HCI Mesh.  See Figure 4 in this vSAN use case to see how HCI Mesh could help achieve this result.

Recommendation:  If you are interested in encryption, choose vSphere's VM Encryption, or vSAN Data-at-Rest Encryption, but not both, as it provides no additive benefit, and will use more host resources and reduce opportunistic space efficiency capabilities.  

Compute and Performance Impacts

Enterprise storage arrays that offer encryption capabilities will perform this process using the CPU processing on the array controllers.  Note that these encryption capabilities are generally limited to "at-rest" encryption.  Three-tier architectures using storage arrays typically do not encrypt the storage fabric that transports the storage traffic.

VMware vSAN was designed with the intention of being able to use any host with hardware that is certified on the vSAN Compatibility Guide. To encrypt and decrypt data efficiently, vSAN uses the Advanced Encryption Standard-New Instructions (AES-NI) CPU offloading capabilities provided by current generation server processors. These advanced instruction offloading capabilities have been present in both Intel and AMD server processors for several years. By offloading encryption tasks through the use of AES-NI processor capabilities, vSAN can easily accomplish encryption with minimal additional overhead to vSphere hosts.  To better understand the additional overhead on hosts and how it may or may not affect VMs, see the post:  Performance when using vSAN Encryption Services.

Device Loss or Theft

Data encryption that occurs at a device level (e.g. self encrypting drives) protects against the physical theft or loss of the device that contains the virtual machine's data. Loss can occur from intentional drive theft, but does not protect from powering off a virtual machine, or cloning the virtual machine, and then downloading that virtual machine to a USB or other portable media device from an administrative console. This is because the data is only encrypted on the underlying storage device, not the storage construct that is presented (such as a block device/LUN or NFS file system). 

When using in-guest encryption solutions, or when using an alternative native VMware encryption solution like VM Encryption, the contents of the virtual machine are encrypted. Data is still secure in the device loss or theft scenario, in addition to protection from downloading a virtual machine to a USB or other portable media device from an administrative console.

While storage devices with "self-encrypting" capabilities exist on the VMware compatibility guide (VCG) for vSAN, it is not supported to run those devices with that feature enabled.  All encryption is required to be performed by vSphere/vSAN.

vSAN Data-at-Rest Encryption

vSAN's Data-at-Rest Encryption service provides encryption for all data objects on a vSAN datastore.  It is a per-cluster setting that provides for prescriptive security when and where you need it.  It is supported on All-flash and hybrid vSAN clusters, and does not require any self-encrypting drives.  

vSAN Data-at-Rest Encryption

The Data-at-Rest Encryption works with all other vSAN features, minimizing any potential operational impact.

Key Management using External KMS

Key management can be accomplished using a cryptographic key provider appliance called a Key Management Server (KMS). KMS solutions provide standards-compliant lifecycle management of encryption keys.  Tasks such as key creation, activation, deactivation, and deletion of encryption keys are performed by Key Management Servers. The  Key Management Interoperability Protocol  (KMIP) can be used to communicate with a KMS by clients to use keys managed by the KMS.

The Domain of Trust, and Trust Establishment 

Three elements comprise a vSAN Encryption domain of trust.  The key provider (either an external KMS, or vSphere NKP), the vCenter Server, and the vSAN hosts in the cluster with encryption enabled.

vCenter Server and vSphere hosts can only use a key provider after establishing a trust with the key provider. Setting up the domain of trust follows the standard Public Key Infrastructure (PKI) based management of digital certificates. A digital certificate must be provided to the KMS from the vCenter Server environment. Different implementations of KMS allow for different types of certificates to be used to establish the trust. These are:

  • Root CA Certificate.  Trust is established for all certificates signed by the root certificate
  • Certificate.  The vCenter Server certificate is used to establish trust
  • New Certificate Signing Request.  vCenter Server generates a CSR, which is submitted to the KMS for signing. The generated certificate is then used to establish trust for the vCenter server.
  • Upload a certificate and private key.  vCenter Server is trusted after the KMS solution's certificate and private key are provided

Once the trust is established between the key provider and vCenter Server, a vSAN cluster (with vSAN Enterprise Licensing) may use vSAN Encryption Services.  When vSAN Data-at-Rest Encryption is enabled on a cluster, the key provider connection information and is pushed to the vSAN hosts. The vSAN hosts then provide a reference key, or key ID, to the key provider. The key provider (in this case, an external KMS cluster) then provides the Key Encryption Key (KEK) that is associated with the key id to the vSAN hosts. Disk keys (DEKs) are wrapped by the the KEK.

Key Management

Key Management tasks as they relate to vSAN Encryption 

Keys are not automatically created for clients even though trusted communication has been established. vSphere administrators do not have direct access to the lifecycle of encryption keys, but actions performed through the vSAN UI impact the key lifecycle process.

  • Turn on Encryption.  Unique KEK is created for each vSAN cluster. DEKs are created when claiming vSAN cache and capacity devices. This is a rolling process.
  • Turn off vSAN Encryption.  DEKs and the KEK are removed from the vSAN cluster in a rolling process.
  • Shallow ReKey.  An existing KEK is recreated for the cluster a shallow rekey is being performed on.
  • Deep ReKey.  The existing KEK and DEKs are recreated. Like enabling or disabling vSAN Encryption, this is a rolling process.

Access to these capabilities can be restricted from vSphere Administrators by assigning user accounts to the No cryptography administrator role.

KMS Availability 

When designing any environment, services such as Domain Naming Service (DNS) or Network Time Protocol (NTP) are typically highly available. Just as DNS is critical for name resolution and NTP is critical for time synchronization, key provider services are critical to the availability of encrypted data. Some Key Management Server vendors provide highly available configuration capabilities, often in the form of configuring multiple KMS Servers into a KMS Cluster. Choosing a key provider solution that provides a resilient and available key provider infrastructure is an important part of the vSAN Encryption design.

An alternative or additional level of resilience can be achieved through key persistence on the hosts through a TPM device installed one each host in the cluster.  More information on TPMs can be found in the Key Persistence section of this document.  

Recommendation:  VMware recommends the use of Trusted Platform Modules (TPM) on each server.  This will allow for the keys distributed to the hosts to be securely stored on persistent media (the TPM) on each host and provide the key to the host when the key provider is inaccessible.

KMS Compatibility 

While there are different KMIP protocol versions available today, VMware vSAN, and VM Encryption, support the  KMS 1.1 protocol. Any KMIP 1.1 compatible Key Management Server solution that provides KMIP 1.1 is supported with vSAN and VM Encryption. 

KMS Infrastructure Placement 

If using one or more KMS virtual appliances, they should not be deployed on an encrypted datastore if the hosts do not have TPMs to persist the keys. This is because placing a KMS appliance/cluster on top of the datastore it is providing keys for, creates a circular dependency. Consider the following unsupported scenario where A KMS appliance or cluster resides on the encrypted cluster it is providing keys for, with hosts that do not have TPMs for persistent key storage.

  • If a single KMS appliance resides on the cluster, if the host it is running on fails/reboots, when the host comes online, it cannot mount the encrypted disks until the KMS returns to service, because the KMS is not available.
  • If a KMS cluster resides on an encrypted cluster and all hosts suffer a power loss, when they are powered on, they will not be able to mount their disks, because the KMS cluster is not available.

Assuming the hosts in the examples above are not using TPMs, in both of these cases, the KMS appliance/cluster will not be available because its storage is not available.

Recommendation:  VMware recommends the use of Trusted Platform Modules (TPM) on each server.  This will allow for the keys distributed to the hosts to be securely stored on persistent media (the TPM) on each host and provide the key to the host when the key provider is inaccessible.

 VMware does not support placing a KMS appliance or cluster on the encrypted datastore it is providing key management for. 

 

Key Management using Native Key Provider (NKP)

vSphere 7 U2  introduced the vSphere Native Key Provider, a mechanism to enable vTPM, VM Encryption, and vSAN Encryption that exists completely within vSphere itself. It is driven by vCenter Server and clustered ESXi hosts and, to vSphere, enables nearly the same functionality as with a traditional Key Management Service (KMS). With this, customers of all sizes have better access to encryption technologies. Additional information can be found at the native key provider homepage.

vSphere NKP

When using the vSphere NKP, many of the principles of key management are similar to using an external KMS solution.  Additional configuration and operational guidance for using the vSphere NKP with vSAN is provided later on in this document.

Key Persistence

Recent editions of vSphere and vSAN support the use of TPM chips to persist the keys safely and securely on each vSAN host to ensure availability of keys in the event that the key provider is inaccessible.  This capability will work for external KMS solutions, and the vSphere NKP.  While it is especially ideal for environments with limited connectivity to a KMS, or non-redundant KMS solutions, the addition of TPM chips on any new vSphere host should be a part of an organization's purchasing practices.  It is a simple and affordable way to make the distribution of encryption keys to vSAN hosts more robust.

TPM

In vSAN 7 U3, when using TPM 2.0 chips on all vSAN hosts in a cluster, any key issued (from a third party KMS or the vSphere NKP) that that is stored in the key cache, it will also be persisted to the TPM chip immediately.  Upon reboot of the host, this key persistence feature would restore the key from the TPM chip to the key cache.  If the KEK, or the host key are not within the key cache for some reason, and cannot be fetched from the TPM, then they will be retrieved from the KMS. Keys will always be attempted to be fetched from the key cache first.  Even upon host restart, the TPM will restore the keys immediately.  If their are no keys locally, then they will be retrieved from the KMS.

Key persistence is enabled by default when using the vSphere NKP, but when using an external KMS solution, will require enabling it through the following esxcli commands.

esxcli system settings encryption set --mode=TPM

esxcli system security keypersistence enable

The use of key persistence through TPM chips only applies to key persistence directly on the vSAN hosts.  Key management of vSAN is limited to interaction of the key providers and the vSAN hosts.  At this time, vSAN does not support key management via attestation through a vTA cluster.

Note that many operational activities related to key management do not change with the use of cached keys on TPM chips in the vSAN hosts.  Examples of operational activities that are unchanged from this enhancement include:

  • Turn on/off encryption (stale/old KEK and old DEK automatically removed from persistent storage on TPM)
  • Shallow rekey (stale/old KEK and old host key automatically removed from persistent storage on TPM)
  • Deep rekey (stale/old KEK, old DEK, and old host key automatically removed from persistent storage on TPM)
  • I/O with data-at-rest encryption.
  • Remove disk (stale/old DEK automatically removed from persistent storage on TPM

Recommendation:  VMware recommends the use of Trusted Platform Modules (TPM) on each server.  This will allow for the keys distributed to the hosts to be securely stored on persistent media (the TPM) on each host and provide the key to the host when the key provider is inaccessible.

vCenter Server

An environment's vCenter Server is the primary management application for vSphere. It is a familiar platform that acts as a control plane for the configuration and management of the different parts of vSphere and additional VMware solutions. vCenter provides a rich set of APIs used by other VMware solutions, 3rd party solutions, and often custom code. VMware solutions use these APIs to provide a uniform and consistent framework for better interoperability and management. 

KMS Configuration

vCenter Server provides a central location for KMS configuration that is available to be used by either vSAN Encryption or VM Encryption. It is also the primary interface when using the vSphere NKP.

KMS Configuration

Certificates used to establish the trust with the KMS are persisted into the VMware Endpoint Certificate Store (VECS). These certificates are shared by both vSAN Encryption and VM Encryption. To ensure proper trust between the hosts and the KMS, certificates and the KEK_ID are pushed to vSphere hosts for vSAN Encryption. Using the KEK_ID and KMS configuration, hosts can directly communicate with the KMS cluster without the dependency of vCenter being available.

vSAN Encryption Configuration

VSAN cluster configuration is often performed within the vSphere Web Client. VSAN is configured per vSphere cluster, and vSAN Encryption is a configuration option of a vSAN cluster. Provided a KMS has been configured, vSAN Encryption is easily enabled through the cluster management UI in the vSphere Web Client.

vSAN Encryption Services

When encryption is enabled a few options are available

  • Wipe residual data (formerly Erase disks before use) - Useful for disks that already have data on them. This wipes any data from the disk before encryption occurs.
  • Allow Reduced Redundancy - vSAN will reduce the protection level during when the service is turned off or on. This is typically only used when a vSAN cluster is at the maximum number of hosts or fault domains required to meet a protection policy.

Securing Administrator Access to Cryptographic Operations

Access to the cryptographic properties and actions of a vSAN cluster, similar to encrypted virtual machine properties/actions for VM Encryption, are limited to users or groups assigned with the Administrators role. It is often necessary to create custom roles or provide "Administrator-like" access to vCenter, vSphere hosts, and the vSAN cluster itself. 

Users assigned to this role may not perform Cryptographic operations in vSphere and vSAN environments.

Roles

 

Operations not associated with cryptographic tasks can be performed by users assigned to this role.   The summary below describes the basic capabilities of this role.

Operations not allowed Operations allowed
Manage KMS Add a host to an encrypted vSAN cluster
Manage encryption policies Move VM from a non-encrypted datastore to encrypted datastore
Manage keys Move VM from a non-encrypted vSAN cluster/datastore to an encrypted vSAN cluster/datastore
Register a VM  
Register a host  

vSAN Hosts

The following described noteworthy considerations about the hosts in a vSAN cluster when vSAN Data-at-Rest Encryption is enabled.

Configuration

When vSAN Encryption is enabled, several items are configured/pushed to the vSphere host. Items such as the KMS Cluster information, the Key Encryption Key ID (KEK_ID), and a host key that is unique for the cluster.

KMS information that is pushed to each host in the vSAN cluster

  • Cluster ID/name of the KMS cluster
  • KMS Port - This is typically 5696
  • Address of the cluster
  • Proxy Address/Port if used
  • A host key used core dumps.

vSAN Encryption specific information that is pushed to/created for the cluster

  • The Key Encryption Key - Used to retrieve the KEK from the KMS on boot up so vSAN Disk Groups can be mounted so data may be read from or be written to them.
  • vSAN Host Key ID - Key used to encrypt/decrypt host core dumps.

Depending on the state of vSAN encryption, some other values are set

  • Whether the host is going through the process of enabling or disabling encryption
  • Whether the current KMS Server is the current, or previous KMS Cluster during a KMS Cluster changing process

Host behavior at boot up

When vSAN Encryption is enabled, to participate in data operations requiring data encryption/decryption, a host must have access to the KEK. Hosts connect directly to the key provider over the Management VMkernel interface to securely retrieve the KEK using the KEK_ID (which is pushed by vCenter Server when enabling vSAN encryption). If the hosts do not use TPMs, the KEK is not persistently stored, but rather stored in secure location in host memory in the key cache kernel module. This kernel module caches keys for allowed processes, and is used by both vSAN Encryption and VM Encryption.

Because the KEK is not persistent, each time a host boots, it must use the KEK_ID and KMS settings to connect to the KMS Cluster and retrieve the KEK. The KEK is then placed in the key cache kernel module for use by vSAN Encryption. With KEK_ID and KMS settings being persistently stored on each host, there is no requirement to communicate with vCenter server to retrieve the KEK. This is advantageous in the situation where vCenter Server may be offline from a failure, reboot, or network isolation.

Recommendation:  VMware recommends the use of Trusted Platform Modules (TPM) on each server.  This will allow for the keys distributed to the hosts to be securely stored on persistent media (the TPM) on each host and provide the key to the host when the key provider is inaccessible.

The boot process for vSAN hosts participating in an encrypted vSAN cluster is as follows:

  1.  The ESXi host boots and once hostd starts, the system appears to be available in vCenter, but it isn't just yet.
  2.  vSAN tells hostd to turn off all core dumps on the host
  3.  vSAN requests a unique key from the KMS for the purpose of encrypting the host core dumps
  4.  vSAN tells hostd to store the host key in the kernel key cache
  5.  vSAN passes the KEK_ID to the KMS
  6.  The KEK retrieved it placed into the kernel key cache, which is used to mount vSAN disks
  7.  vSAN mounts the encrypted disks.

This sequence of events differs from the normal boot process, as vSAN disks are mounted before hostd becomes available. Because of this process, when vSAN Encryption is enabled, it is possible to have a vSAN host "up" but still performing the process of mounting vSAN disks. 

*If a running host is added to an encrypted vSAN cluster, it will not immediately have access to the existing vSAN datastore. A KEK request must be performed. This can occur by creating disk groups on the newly added host, or it may be requested from a reboot of the newly added host.

Note that for stretched cluster and 2-node topologies, the vSAN witness host appliance is not encrypted.  Why?  It is more secure to not encrypt the witness node. If the witness node is encrypted, it has to store all the credentials to get the secret key from the Key Management Server (KMS). These credentials become another attack surface that we have to protect. However, since witness runs in a virtual environment, it is easier to be attacked than regular hosts which run in physical environments. Not encrypting witness node reduced attack surface and makes the system more secure.  What can be leaked on the witness node includes number and size of each vSAN object, their log sequence number, and policy. None of these are sensitive user data.

vSAN Disk Groups

Data-at-Rest Encryption occurs within each device that is part of a disk group.  Every time vSAN Data-at-Rest Encryption is enabled or turned off, each disk group in the vSAN cluster goes through a Disk Format Change (DFC). When enabling encryption, a new partition is added that holds a small amount of meta-data used by vSAN to manage operations on the encrypted cluster. This step essentially prepares the disk to encrypt any write that is directed to it. 

A "generation-id" is created the 1st time encryption is enabled. Each time encryption is then turned off, reenabled, or a Deep ReKey is performed, the generation-id increments by a value of 1. The DFC process evaluates the generation-id for each device in a disk group to determine if a DFC needs to occur on that disk. This is especially beneficial in cases where the DFC process is interrupted or in cases where hosts were offline during a Deep ReKey.  The following shows the disk configuration before and after vSAN Data-at-Rest Encryption is enabled.

DFC change

 

When Disk Format Change operations take place, it occurs as a rolling upgrade: one disk group at a time.  If data is present anywhere in the disk group, the data is evacuated to ensure data is preserved. 

An optional feature of "Erase disks before use” will soft erase the disk before writing new data, but should be used with care, as it can be time consuming. It is largely unnecessary unless the device is being decommissioned or returned to the manufacturer.  For more information on device erasing considerations, see the section "Guidance when using 'Erase disks before use."

Writing data to an encrypted vSAN datastore

Encryption occurs as the last step on an I/O flow, for the highest level of protection and efficiency of deduplication and compression.

Encryption data path

In order to continue to provide the benefits of deduplication and compression to encrypted clusters, the data must be handled by vSAN in a specific order.  For vSAN clusters using Data-at-Rest encryption and Deduplication and Compression services, the order is as follows:

As data is written to to the write buffer/caching tier: 

  1. Write I/O broken into 64K chunks
  2. Checksum performed on 4K blocks
  3. Encryption performed on 4K blocks
  4. Lands in the write buffer

As data is destaged from the write buffer to the capacity tier:

  1. Decryption is performed on 4K blocks
  2. Deduplication is performed on 4K blocks
  3. Compression is performed on 4K blocks
  4. Encryption is performed on 2-4K blocks
  5. Lands in the capacity tier

When using vSAN Data-at-Rest Encryption, it will only encrypt the data at rest.  To encrypt the data in flight, enable the vSAN Data-in-Transit service on the cluster.

Role Based Access Control

Securing workloads does not end with the use of encryption technologies. Access granted to data and its management must also be properly secured. Effective access to these workloads must align with the responsibilities associated with their management, configuration, reporting, and user requirements.

No Cryptography Administrator Role

The "No Cryptography" role is very similar to the normal administrator with many of the same privileges. Operations such as power on or off a virtual machine, boot, shutdown, vMotion, as well as normal vSAN management may be performed. However, this role is not allowed to perform any cryptographic operations.

No Cryptography Administrator Role

The permissions in the illustration, show that users assigned the No Cryptography Administrator role do not have any permissions to perform any operations that require any cryptographic operations.

No Cryptography Administrator and VM Encryption

Users assigned to the No Cryptography Administrator role are not granted the following privileges:

  • Ability to encrypt or decrypt virtual machines with VM Encryption
  • Direct console access to virtual machines that are encrypted with VM Encryption
  • The ability to download virtual machines that are encrypted with VM Encryption. This will prevent the user from downloading a virtual machine to a USB or other offline media.
  • The ability to add hosts to vCenter. This limitation exists, because the process of adding a host to vCenter grants the host access to the cryptographic keystore.

No Cryptography Administrator and vSAN Encryption

Users assigned to the No Cryptography Administrator role are not granted the following privileges:

  • The ability to enable or turn off vSAN Encryption
  • The ability to generate new encryption keys (Shallow or Deep Rekey)
  • The ability to add hosts to vCenter.

Users assigned to the No Cryptography Administrator role are granted the following privileges:

  • Direct console access to virtual machines that reside on a vSAN Cluster with vSAN Encryption enabled
  • The ability to download virtual machines that reside on a vSAN Cluster with vSAN Encryption enabled.
  • The ability to add hosts to a vSAN Cluster*.

* In a situation where a host needs to be added to a vSAN Cluster, a user with Cryptographic rights would have to add the host to vCenter. Once added to vCenter a Non-Cryptographic Administrator could then add the host to an encrypted vSAN Cluster.

vSAN Data-in-Transit Encryption

Complimenting vSAN Data-at-rest encryption, vSAN also provides a cluster-based feature for encrypting vSAN data in transit.  The vSAN “Data-in-Transit Encryption” securely encrypts all vSAN traffic in transit across hosts.  It uses the same FIPS 140-2 validated Cryptographic modules as used with Data-at-rest encryption, as well as vSphere’s VM Encrypt, and does so in an automated manner that does not require a KMS server for key management.  Data-in-Transit Encryption option can be enabled independently from the vSAN data-at-rest encryption, but enabling both will provide a complete, end-to-end encryption solution on a per-cluster basis.

Data-in-Transit Encryption

Unlike traditional three-tier architectures, where the storage fabric may be physically isolated from all other types of traffic, vSAN may use the same uplinks to serve multiple needs.  vSAN Data-in-Transit Encryption addresses this architectural difference and provides a capability that is typically not found in other architectures: over-the-wire encryption. 

The feature is compatible with most data services offered by vSAN, such as deduplication and compression, as well as compression-only.  It is not supported with HCI Mesh at this time.

Since there are no management tasks related to key management for vSAN Data-in-Transit Encryption, the majority of guidance in this document is focused on vSAN Data-at-Rest Encryption.

Key Management

All key management tasks are handled automatically by the vSAN hosts participating in the cluster, which means that there is no need for a key provider (external KMS or vSphere NKP) if vSAN Data-in-Transit is enabled. If both encryption services are enabled, Data-at-Rest Encryption will use it's own built-in key provider (e.g. external KMS or vSphere NKP), while the Data-in-Transit encryption will continue to manage the keys for Data-in-Transit encryption automatically.

When Data-in-Transit Encryption is enabled, all hosts that join a vSAN cluster are authenticated with dynamically generated symmetric keys.  Upon the removal of a host from a cluster, any existing authentication is removed.

vSAN Encryption Services Operations

The following will detail many of the common operations related to vSAN clusters that use vSAN encryption services.  Other operational recommendations for vSAN Encryption Services can be found in the "Data Services" section of the vSAN Operations Guide.

 

Enable vSAN Encryption

Enabling vSAN encryption services is as easy as clicking on one toggle in vCenter Server.  Before attempting to enable vSAN Encryption, a few items need to be considered.

  • Is AES-NI supported and enabled in each the bios of each host in the vSAN cluster? The encryption process takes advantage of Advanced Encryption Standard New Instructions (AES-NI) in many of today's CPU's. These additional instruction sets supported by AES-NI enabled processors, perform much of the work of the encryption and decryption, removing the need to perform these tasks in software alone. Some server configurations have AES-NI enabled by default, while others do not. Consult the system manufacturer's documentation to determine whether AES-NI is supported and how to verify it is enabled.
  • What is the state of Deduplication & Compression?  The process of enabling or disabling vSAN Encryption requires a disk group format change. This is performed in a rolling process through the cluster.  Changing the state of deduplication & compression also requires a disk group format change, and is also performed in a rolling process through the cluster. Enabling/disabling encryption and deduplication & compression independently would require multiple disk group format changes.  As a result, it may be better perform encryption and deduplication & compression setting changes simultaneously. This allows for both setting changes to occur during one disk group format change in a rolling process through the cluster.
  • Will Reduced Redundancy be required?  Because the process of enabling or disabling encryption requires a disk group format change, there must be enough nodes or fault domains for the data being moved off of a disk group to reside elsewhere in the vSAN cluster.  3-Node vSAN configurations will require Reduced Redundancy because there is no additional node for the data on disk groups being evacuated to move to. This may not be the case in a 3 fault domain configuration, depending on how many nodes are in each fault domain.  2-Node vSAN configurations are treated identically to 3 node vSAN configurations when performing operations requiring an disk format change. Reduced Redundancy will be required in 2 Node vSAN configurations as well.  Configurations that have a sufficient number of fault domains may also require Reduced Redundancy in situations where there is not enough free capacity to migrate data being evacuated from a disk group. 

vSAN Encryption services can have an impact on performance.  For a more detailed understanding, see the blog post:  "Performance when using vSAN Encryption Services."

Permissions required to enable vSAN Encryption

To enable vSAN encryption, user or group first must have the proper access. A user or group must have the following permissions:

  • Host > Inventory > Modify Cluster
  • Cryptographic Operations > Manage encryption policies
  • Cryptographic Operations > Manage KMS
  • Cryptographic Operations > Manage keys

Enabling vSAN Encryption 

Select Cluster > Configure > Services > Edit to enable vSAN Encryption

Enable Encryption

If the user or group does not have the proper permissions, the Encryption option will not be presented.  Referring to the suggestions previously covered, consider the options:

  • Deduplication and Compression
    • If there is no long term desire to change the state, then do not change this setting. 
    • If planning to change this setting in the near term, it may be better to make this change at the same time as changing encryption.
  • Encryption - Select to enable
    • This will force a disk format change from the current state.
    • KMS connectivity and cluster key ID information is pushed to the vSAN hosts to enable the ability to communicate directly with the KMS server.
    • A host key is created for the cluster
  • Wipe residual data (formerly Erase disks before use)
    • This will wipe any existing data from a disk as the encryption process occurs
    • When wiping any existing data from a disk the encryption enablement process has a significantly longer duration
    • Any time additional disks are added to a disk group or a disk group is created drives will be wiped before being added to the disk group
  • Allow Reduced Redundancy - Will be required if the conditions previously covered regarding 2 Node, 3 Node, or there is minimal available capacity preventing a disk group evacuation.

Select Ok

The vSAN Encryption process will begin.

Turning off vSAN Encryption

Turning off vSAN encryption services is as easy as clicking on one toggle in vCenter Server.  Before attempting to turn off vSAN Encryption, a few items need to be considered.

  • What is the state of Deduplication & Compression?  The process of enabling or disabling vSAN Encryption requires a disk group format change. This is performed in a rolling process through the cluster. Changing the state of deduplication & compression also requires a disk group format change, and is also performed in a rolling process through the cluster. Toggling a status change of encryption and deduplication & compression independently would require multiple disk group format changes.  As a result, it may be better to perform encryption and deduplication & compression setting changes simultaneously. This allows for both setting changes to occur during one disk group format change in a rolling process through the cluster.
  • Will Reduced Redundancy be required?  Because the process of enabling or disabling encryption requires a disk group format change, there must be enough nodes or fault domains for the data being moved off of a disk group to reside elsewhere in the vSAN cluster.  3-Node vSAN configurations will require Reduced Redundancy because there is no additional node for the data on disk groups being evacuated to move to. This may not be the case in a 3 fault domain configuration, depending on how many nodes are in each fault domain.  2-Node vSAN configurations are treated identically to 3 node vSAN configurations when performing operations requiring an disk format change. Reduced Redundancy will be required in 2 Node vSAN configurations as well.  Configurations that have a sufficient number of fault domains may also require Reduced Redundancy in situations where there is not enough free capacity to migrate data being evacuated from a disk group. 

Permissions required to change the status of vSAN Encryption Services

To change the status of vSAN encryption services, user or group first must have the proper access. A user or group must have the following permissions:

  • Host > Inventory > Modify Cluster
  • Cryptographic Operations > Manage encryption policies
  • Cryptographic Operations > Manage KMS
  • Cryptographic Operations > Manage keys

Turn off vSAN Encryption 

Select Cluster > Configure > General > Edit to turn off vSAN Encryption

Disable Encryption

If the user or group does not have the proper permissions, the Encryption option will not be presented.  Referring to the suggestions previously covered, consider the options:

  • Deduplication and Compression
    • If there is no long term desire to change the state, then do not change this setting. 
    • If planning to change this setting in the near term, it may be better to make this change at the same time as changing encryption.
  • Encryption - Deselect to turn off
    • This will force a disk format change from the current state.
    • Encryption will be turned off for the cluster
  • Allow Reduced Redundancy - Will be required if the conditions previously covered regarding 2 Node, 3 Node, or there is minimal available capacity preventing a disk group evacuation.

Select Ok

The process of turning off vSAN Encryption will begin.

External KMS Operations

When using vSAN Data-at-Rest Encryption, a KMS must be used to manage the keys used to encrypt and decrypt the data.  The KMS can either be an external third party KMS solution, or the vSphere Native Key Provider (NKP).  vSAN Data-in-Transit encryption will always manage its own keys, and thus does not need or use a key provider, even if it is used in combination with the Data-in-Transit Encryption service.

The process of configuring a KMS server is relatively simple. The process can be broken down to these steps:

  • Add the KMS
  • Have the KMS trust vCenter Server.

Adding the KMS

To add the KMS, select vCenter in the Navigator panel, choose Configure, then select Key Management Servers.

Adding the KMS

If the Key Management Servers option isn't visible, the account logged in does not have Manage KMS privileges in vCenter Server.

Manage KMS privileges in vCenter

To add a KMS, select Add KMS and add the KMS cluster properties. Depending on the network configuration, a proxy and credentials may be required. The KMS Server port will normally be 5696, but an alternate port may be used.

The KMS Server port will normally be 5696

When asked to Trust the KMS Server's certificate, select Trust.

Trust the KMS Server's certificate

At this point, vCenter Server trusts the KMS server.

Have the KMS trust vCenter

For the KMS server to properly communicate with the vCenter Server, and ultimately vSAN hosts, the trust has to be bidirectional. 

Have the KMS trust vCenter

To establish the trust with the KMS, select Establish trust with KMS

One of four options will be available:

  • Root CA Certificate
  • Certificate
  • New Certificate Signing Request
  • Upload certificate and private key

Consult the vendor-specific documentation for the proper trust establishment process for the chosen KMS provider.

As an example of establishing trust using a certificate and private key, select  KMS certificate and private key

select  KMS certificate and private key

When prompted, either upload the certificate file and private key file, or past the contents and select Ok.

either upload the certificate file and private key file

When the certificate and private key have been uploaded, the trust will be established.

the certificate and private key

Native Key Provider Operations

Creating a Native Key Provider

Add Native Key Provider

A new native key provider can be configured in a few simple clicks.

SetupNKP

 

Configuring vSAN to use a configured Native Key Provider

Selecting a native key provider is done from the existing vSAN cluster configuration settings.
 

Native Key Provider Design Considerations

  • Native Key Provider is not a full featured KMS. It cannot be used for other KIMP needs.
  • vSAN Encryption requires vSAN Enterprise, or Enterprise Plus licensing.

Backing up native key provider keys

A manual backup of the key provider should be performed from withing the vCenter UI. Do note that when you configure the Native Key Provider and want to back it up, you need to access the vSphere UI via the fully qualified domain name. vSphere Native Key Provider is backed up as part of the vCenter Server file-based backup. However, you must back up the vSphere Native Key Provider at least once before you can use it. When you create a vSphere Native Key Provider, it is not backed up.

Screen Shot 2021-05-24 at 1.55.30 PM

Enforce usage of vSAN Encryption

Additional SPBM storage rules can specify that a virtual machine or disk being created will be placed on a Data-At-Rest encrypted cluster.

image 79

 

Changing KMS Server

The process of changing KMS Servers is essentially a Shallow Rekey operation.

Changing the KMS Server

  1. The initial KMS configuration is in place
  2. The administrator selects an alternate KMS Cluster
  3. The new KMS configuration is pushed to the vSAN hosts
  4. A new host key is generated
  5. vSAN performs a Shallow Rekey

To also perform a Deep Rekey, this should be accomplished after the initial Shallow ReKey has taken place.

Changing the KMS Server via PowerCLI

The process of changing KMS Servers is essentially a Shallow Rekey operation but designating a new KMS Server. The Start-VsanEncryptionConfiguration cmdlet can be used to change the KMS.

Change the KMS with PowerCLI:

Start-VsanEncryption -Cluster "ClusterName" -KMS "KMS Profile Name" -Confirm:$False (Confirm false to proceed without prompting)

 

Secure Device Wiping Options

The "Wipe residual data" (formerly Erase disks before use) is an option available in the vSAN Services dialog box.

Guidance when using Erase disks before use

When a vSAN cluster has this setting checked, random data is written to each device in a disk group before it is encrypted.  Using random data helps ensure a fully secured wipe, as only ones or only zeros may not have the desired effect when using deduplication and compression.

This process can be very time-consuming.  The performance characteristics of the media type, bus protocol, capacity, and number of devices will all impact the time that it takes to complete this operation.

This wipe process aligns with the NIST 800-88 Revision 1 “Clear” definition:

Clear applies logical techniques to sanitize data in all user-addressable storage locations for protection against simple non-invasive data recovery techniques; typically applied through the standard Read and Write commands to the storage device, such as by rewriting with a new value or using a menu option to reset the device to the factory state (where rewriting is not supported).

Further inspection of NIST 800-88 Revision 1specifically Appendix A, addresses different requirements for different media types when it comes to a Clear operation. The following table is a summary relevant to vSAN device types:

Media Type Guidance for wipe operations
ATA Hard Disk Drives PATA, SATA, eSATA, etc Overwrite media by using organizationally approved and validated overwriting technologies/methods/tools. The Clear pattern should be at least a single write pass with a fixed data value, such as all zeros. Multiple write passes or more complex values may optionally be used.
SCSI Hard Disk Drives Parallel SCSI, SAS, FC, UAS, and SCSI Express Overwrite media by using organizationally approved and validated overwriting technologies/methods/tools. The Clear procedure should consist of at least one pass of writes with a fixed data value, such as all zeros. Multiple passes or more complex values may optionally be used.
ATA Solid State Drives (SSDs) PATA, SATA, eSATA, etc 1.  Overwrite media by using organizationally approved and tested overwriting technologies/methods/tools. The Clear procedure should consist of at least one pass of writes with a fixed data value, such as all zeros. Multiple passes or more complex values may alternatively be used.Note: It is important to note that overwrite on flash-based media may significantly reduce the effective lifetime of the media and it may not sanitize the data in unmapped physical media (i.e., the old data may still remain on the media).
2.  Use the ATA Security feature set’s SECURITY ERASE UNIT command, if supported.
SCSI Solid State Drives (SSSDs) Parallel SCSI, SAS, FC, UAS, and SCSI Express Overwrite media by using organizationally approved and tested overwriting technologies/methods/tools. The Clear procedure should consist of at least one pass of writes with a fixed data value, such as all zeros. Multiple passes or more complex values may alternatively be used.
Note: It is important to note that overwrite on flash-based media may significantly reduce the effective lifetime of the media and it may not sanitize the data in unmapped physical media (i.e., the old data may still remain on the media).
NVM Express SSDs NVMe devices Overwrite media by using organizationally approved and tested overwriting technologies/methods/tools. The Clear procedure should consist of at least one pass of writes with a fixed data value, such as all zeros. Multiple passes or more complex values may alternatively be used.

The “Wipe residual data” option meets the requirements of each of these. Though all zeros are not being written, random data (a more complex value) is written instead.

A secure wipe feature is available for vSAN 7 U1 and newer that will allow for  a securely erased device per NIST standards.  For an example in how to use these PowerCLI command, see the blog post:  vSAN - A Secure Fortress for your Data.

Secure disk wipe

The process of enabling vSAN Encryption only encrypts new data. Whether it is an existing cluster, or simply an existing host being added to a vSAN cluster, any residual data could potentially still be recovered. This wiping process ensures there is no residual data on a storage device used by vSAN.

Recommendations for “Wipe residual data” when using vSAN Encryption are:

  • Select “Wipe residual data”
    • When enabling vSAN Encryption for existing vSAN clusters that have vSAN objects on them
    • When adding a host that has data on local devices to an encrypted vSAN cluster
    • When performing a rekey operation to invoke a deep rekey (requesting a new KEK and new unique DEKs created for each vSAN storage device)
  • Deselect “Wipe residual data”
    • When enabling vSAN Encryption for a new vSAN cluster that has not previously had data on the vSAN devices
    • When adding a host that has not had data on local devices that is being added to an encrypted vSAN cluster
    • When performing a rekey operation to invoke a shallow rekey (only requesting a new KEK)

Replacing vCenter when vSAN Encryption is enabled

In order to recover from this and similar scenarios, it is necessary to create a new cluster with the same exact configuration that was originally in use by vSAN Encryption.  The same KMS must be used, as well as have the same KMS Cluster ID.  It is imperative that the same KMS cluster ID remains in order for the recovery feature to work. 

Although the old vCenter Server is gone, the hosts still have the information and keys from the KMS cluster, if we connect to the same KMS cluster with the same cluster ID, the hosts will be able to retrieve the key (assuming the key still exists and was not deleted). The KMS credentials will be re-applied to all hosts so that hosts can connect to KMS to get the keys.

Recommendation:  VMware recommends the use of Trusted Platform Modules (TPM) on each server.  This will allow for the keys distributed to the hosts to be securely stored on persistent media (the TPM) on each host and provide the key to the host when the key provider is inaccessible.

vCenter is lost and KMS information isn't documented

If vCenter is lost, and in cases where the KMS are KMS Cluster ID are not documented, these items can be recreated with a newly deployed vCenter?  In the diagram below we see how the keys are distributed to vCenter, hosts, etc. The KMS server settings are passed to hosts from vCenter by the KEK_id. 

Key provider

In order to obtain the KMIP Cluster ID, we need to look for it under the esx.conf file for the hosts.  You can use cat, vi, or grep (easier) to look at the conf file. You want to look for kmipClusterId, name(alias), etc. Make sure the KMS cluster on the new vCenter configured exactly as it was previously.

cat /etc/vmware/esx.conf 

or something easier…

grep “/vsan/kmipServer/” /etc/vmware/esx.conf

esx.conf

After the KMS cluster has been added to new vCenter as it was configured in the previous vCenter Server, there is no need for reboots. During reconfiguration the new credentials will be sent to all hosts and such hosts should reload keys for all disks in a few minutes.

Shallow Rekey via UI

A Shallow Rekey is performed to change the KEK associated with a vSAN cluster. This is a simple process that can be accomplished from the vSphere Client.

Select the vSAN cluster,  Configure, then  Services, and select  Generate New Encryption Keys 

Shallow Rekey via UI  

This process will create new a new KEK for the cluster and push it to the hosts. Each device's DEK will then be re-wrapped with the new KEK+DEK combination.

Deep Rekey via UI

A Deep Rekey is also a simple process accomplished by vCenter Server. The process is the same as performing a Shallow ReKey, with the  Also re-encrypt all data on the storage using new keys option. This will trigger a Disk Format Change (DFC). A DFC will evacuate data from each device in the disk group in the same fashion as the process of enabling encryption. 

Deep Rekey via UI  

Notice that when a Deep Rekey is selected, the  Allow Reduced Redundancy option is enabled. This should be considered when performing a Deep ReKey as considered when enabling or disabling encryption, depending on the number of available hosts (or fault domains) and available capacity.

Shallow/Deep Rekey via API/PowerCLI

Rekeying is also available using PowerCLI or other vSAN Management API methods.

Shallow/Deep Rekey via API/PowerCLI  

The PowerCLI cmdlet Start-VsanEncryptionConfiguration can perform a Shallow or Deep Rekey. The syntax is as follows:

Shallow Rekey:

Start-VsanEncryption -Cluster "ClusterName" -ShallowRekey -Confirm:$False (Confirm false to proceed without prompting)

Deep Rekey:

Start-VsanEncryption -Cluster "ClusterName" -DeepRekey -AllowReducedRedundancy (if desired) -Confirm:$False (Confirm false to proceed without prompting)

Add Non-Encrypted Host to vSAN Cluster

Adding a new host to an existing vSAN Cluster:

  1. Add the host to the encrypted vSAN Cluster - Compute only nodes or nodes contributing storage to vSAN
    • The hostKey will be installed on the host
    • Configure a VMkernel interface with vSAN Traffic that will allow connectivity with the other hosts in the vSAN cluster
    • The host will be able to access the encrypted vSAN datastore
  2. Add one or more disk groups to the vSAN Cluster - Nodes contributing storage to vSAN
    • When disks are claimed, the KEK will be requested from the KMS cluster, disks will be added after a Disk Format Change occurs. This occurs without rebooting the host.
    • Data may be written to the encrypted disk group(s) on the new host.

vSAN Encryption Troubleshooting

KMS Server Accessibility

The availability of the key provider (whether it be an external KMS, or using the vSphere Native Key Provider in vSphere) plays an important role in the proper distribution and management of keys.  

If the KMS resides on the datastore it is providing key management for, and the hosts do not have TPMs to persist the keys, disk groups will not be mounted if the keys are unavailable.  Hosts in a vSAN cluster that has vSAN Encryption enabled and do not have TPMs to persist the keys, will directly contact the KMS they are assigned to upon boot up to unlock/mount disk groups.

Consider the following scenario:

  1. KMS resides on a vSAN cluster that has vSAN Encryption enabled.
  2. Hosts that have KMS disks for a virtualized KMS appliance lose power. The KMS is then not accessible.
  3. Those hosts are rebooted, and attempt to connect to the (now unavailable) KMS appliance.
  4. The previously failed vSAN hosts will boot, but will not unlock or mount the disk groups.
  5. The KMS appliance’s disks are still not available and will not be.

While it remains advisable to house the KMS appliances in another location (perhaps a management cluster) than the vSAN datastore it is providing keys for, using TPMs in the hosts will ensure that keys will persist in the event that there is an issue with the KMS on the cluster, or anywhere else in the infrastructure. 

Sample PowerCLI code exists that can be used to check and see if a KMS appliance is residing on the vSAN Cluster it is providing key management for located here: https://code.vmware.com/samples/3773/

Recommendation:  VMware recommends the use of Trusted Platform Modules (TPM) on each server.  This will allow for the keys distributed to the hosts to be securely stored on persistent media (the TPM) and ensure keys can be used under these types of failure conditions.

KMS Profile Addressing

When using vSAN Encryption, one of the vSAN Health Check tests will show the health of the connection between the vSAN Hosts and the KMS Cluster as well as vCenter and the KMS Cluster.  A recent scenario came up where the vSAN Health Check indicated that the vSAN Hosts could properly communicate with the KMS Cluster, but the vCenter server had intermittent connectivity to the KMS Cluster.

Troubleshooting indicated that there were no blocked ports between the vCenter Server and the KMS Cluster as well as they were able to properly ping each other. vSAN Hosts could properly ping the KMS Cluster as well, and no ports were blocked.

Here is the vSAN Health Check’s reported error for the vCenter KMS Status.

KMS Profile Addressing

Notice that the certificate status is valid, but the connection and trust statuses are not.

Looking at the Host KMS Status it can be seen that the hosts are properly communicating with the KMS Server.

properly communicating with the KMS Server

The process of enabling vSAN Encryption includes the following steps:

  1. A KMS Connection Profile is created in vCenter and the trust is established.
  2. vSAN Encryption is enabled in the Configuration>Data Services menu in the vSAN UI.
  3. The KMS Connection Profile is pushed to each of the ESXi hosts, they use the kekId and hostkeyId in this profile to retrieve the KEK and HostKey for the vSAN Cluster.

The connection has to be correct in vCenter Server before it can be correct/pushed to vSAN Hosts. Something must have changed in the environment to cause this issue.  Further investigation indicated that the connectivity to the KMS Cluster was intermittent. Sometimes the vCenter KMS Status reported green and other times reported red. So maybe nothing changed.  Careful review of the vCenter KMS Status and Host KMS Status health checks, the KMS Alias is a “short name.”  Maybe there is an issue where the short name is intermittently resolved from DNS… But the vSAN Hosts were not showing any intermittent connectivity, only the VCSA.  The Key Management Servers configuration Profile in the vCenter’s settings shows that the trust cannot be established. The KMS Address is the same value as the KMS Alias in the vSAN Health Check.

When using a short name, the default TCP/IP stack of a vSAN host uses designated search domains in the name resolution process. In the case of this cluster, demo.local and demo.central can be used in short name resolution.

The VCSA, on the other hand, does not have any search domains:

Without search domains to assist with the short name, vCenter Server would rely on the DNS server for name resolution.

The suggestion was made to change the KMS Address value for each KMS Cluster node to either an IP address or the Fully Qualified Domain Name (FQDN). Changing one of the two KMS entries showed some success.

Adjusting the KMS Address for the alternate KMS Cluster node cleared the issue up entirely.

In the case that this was brought up, an alternate vCenter Server had no issues connecting to the KMS Cluster, but an IP address was used instead of a short name. Without digging into DNS configurations of the environment, setting the Fully Qualified Domain Name (FQDN) resolved the issue.

In summary, when configuring the Key Management Server connection profile for a KMS Cluster, ensure that the KMS Address is one that vCenter and vSAN hosts can correctly resolve. Using a Fully Qualified Domain Name or IP address can prevent “short name” related issues.

Booting when vCenter Server is Unavailable

The Host Key and KEK are not stored on vSAN hosts, but rather stored in the key cache after being requested by the vSAN host when vSAN Encryption is enabled. When a vSAN host reboots, these keys are discarded. When a vSAN host reboots, because the Host Key and KEK are not present, they must be requested directly from the Key Management Server if the hosts do not use a TPM, which will persist the keys safely on the host. The Key Management Server profile, Host Key Id, and KEK Id information stored in /etc/vmware/esx.conf is used to request the Host Key and KEK.

Recommendation:  VMware recommends the use of Trusted Platform Modules (TPM) on each server.  This will allow for the keys distributed to the hosts to be securely stored on persistent media (the TPM) and ensure keys can be used under these types of failure conditions.

The example below demonstrates the values that can be found in /etc/vmware/esx.conf for vSAN Encryption:

configuration

When vSAN Encryption is enabled, or when a deep rekey operation is invoked, the vSAN host creates a unique DEK (XTS-AES-256) for each device, and it is encrypted with the KEK. A shallow rekey operation swaps out the KEK and rewraps each DEK.  When a host with vSAN Encryption enabled attempts to mount a vSAN Disk Group, the DEK is unwrapped using the KEK, allowing vSAN to mount and then use the vSAN disk group.

The Boot Process

An encrypted vSAN cluster will demonstrate the following behavior at boot up given the set of conditions below:

  • Entire vSAN encrypted vSAN cluster is offline.
  • vCenter Server residing on offline vSAN cluster is also unavailable
  • Hosts in offline vSAN cluster do not use TPM devices
  • External KMS is online

In the scenario above, once the hosts are powered up, the disks and disk groups are not immediately mounted, and thus the VMs, including vCenter server are offline.

  1. The host boot process will read the values in /etc/vmware/esx.conf and request the KEK and Host Key from the KMS using the KEK Id and the Host Key Id respectively, directly from the KMS.  This is because the previous keys were retained in the non-persistent key cache, and the hosts do not have a TPM to persist the keys. 
  2. The KEK and Host Key are placed in memory in the key cache. If the hosts used TPM devices, they would be cryptographically stored on the TPM devices in each host of the cluster.  At this point the KEK is then used to mount the vSAN Disk Groups.  The VMs, including vCenter Server can be powered on.

As long as vSAN hosts using vSAN Encryption have connectivity to their configured KMS or have them cached on a TPM, they have no issue booting, even when vCenter is offline. The boot process is not dependent on vCenter to unlock and mount vSAN Disk Groups

If the key provider (KMS) were also unavailable in this scenario, then the disk groups would not be mounted and the VMs would be unavailable.  Using TPMs in each host helps avoid this scenario and improve the robustness of an encrypted cluster substantially.

Recommendation:  VMware recommends the use of Trusted Platform Modules (TPM) on each server.  This will allow for the keys distributed to the hosts to be securely stored on persistent media (the TPM) and ensure keys can be used under these types of failure conditions.

 

Appendix A: Common Terminology

Below are common encryption terms throughout this document, and how they pertain to vSAN Encryption Services:

  • Key Provider:  The entity providing keys.  May be referring to either an external third party Key Management Server (KMS), or the vSphere Native Key Provider (NKP)
  • KMIP : Key Management Interoperability Protocol.
    • A standard protocol that clients talk to KMS.
    • The KMIP 1.1 protocol is required for use with vSAN Encryption
  • KMS : Key Management Server.
  • KMS Cluster : A cluster of KMS servers.
    • The servers in the cluster maintain replication (mostly synchronous replication) so every key operation that renders a modification will be reflected by other server nodes immediately.
    • KMS cluster resiliency and availability is paramount to consider when implementing any encryption solution.
  • KEK : Key Encryption Key.
    • This is the key stored in KMS. This is a per-tenant key, resulting in each vSAN cluster having one KEK.
    • Key Encryption Keys are AES-256
  • DEK : Data Encryption Key.
    • This is the key used in the I/O path to encrypt/decrypt data.
    • Data Encryption Keys are XTS-AES-256 keys.
    • Each disk in a vSAN disk group will have a DEK.
  • Host Key : This is similar to KEK, but is used to encrypt vSAN host core dumps, not data.
    • All hosts in a vSAN cluster use the same HostKey.
    • By providing a Host Key, customers can safely send encrypted core dumps to VMware Global Support without disclosing DEKs.
    • This assists in maintaining the integrity of customer data, while assisting VMware Global Support with problem resolution.
    • vSAN Host Keys are AES-256
  • Wrapped : Wrapped is synonymous with encrypted.
    • "X" wrapped by "Y" means the clear text of "X" was encrypted using "Y" as the key, and the "Y" is needed to unwrap the wrapped key.
    • With vSAN Encryption, after the DEK is wrapped using the KEK, it is stored on persistent media.
  • Rekey : change the key used in encryption.
    • Shallow rekey : change the KEK only. The DEK is wrapped with a new KEK. This is usually very fast.
    • Deep rekey : change the DEK for each device and re-encrypt all data using each device's new DEK.
      This will be very slow because all data needs to be rewritten.
  • TPM.  A Trusted Platform Module (TPM) is a device that sits inside of a host, that can cryptographically store keys that are issued to the host from the key provider.  This module is an affordable (typically around $50 USD) way to ensure that host keys can be retrieved when there is an issue with communication to the key provider.  Whether an environment is using the vSphere Native Key Provider (NKP) or a dedicated KMS cluster and vSAN 7 U3 or newer, using a TPM in each host is highly recommended.
  • Key cache : A vSphere Host kernel module that caches the KEK from the KMS for use by vSAN Encryption and VM Encryption.
  • FIPS 140-2 : The Federal Information Processing Standard ( FIPS ) Publication 140-2, is a U.S. Government standard for computer security that is used to approve cryptographic modules. The title is Security Requirements for Cryptographic Modules. It was initially published on 25 May, 2001. More information can be found on the NIST site .

Additional References

  • VMware VM Encryption and vSAN Encryption FAQ
    https://core.vmware.com/resource/vsan-frequently-asked-questions-faq#section14 
  • Blog:  Performance when using vSAN Encryption Services
    https://core.vmware.com/blog/performance-when-using-vsan-encryption-services
  • Blog: Support for Key Persistence
    https://core.vmware.com/blog/support-key-persistence 
  • TPM devices
    https://core.vmware.com/resource/vmware-vsan-design-guide#sec6864-sub4 

Associated Content

From the action bar MORE button.

Filter Tags

Storage vSAN vSAN 6.7 vSAN 7 vSAN Encryption Document Best Practice Deep Dive Advanced Design Deploy Manage Migrate Optimize