Inscreva-se no feed

Worried about the security of your application data at the edge? Encryption at rest provides a solution.

Edge environments are usually located at sites with untrusted networks and lesser physical security than traditional data centers. This puts the security of the application's data running on them at risk. Data destruction, corruption or even the leak of confidential information to malicious actors is possible. To reduce that risk it is recommended to encrypt every disk storing sensitive data, like the one where the application's persistent data is stored.

This article describes how to create an encrypted disk volume during a Single Node OpenShift (SNO) deployment (a SNO deployment has just one node, where the roles of control plane and worker reside in the same machine) and use Logical Volume Manager (LVM) Storage Operator to create a volume group and provision Persistent Volume Claims (PVCs) for your application's data.

Please note: Software encrypted drives are not officially supported by Red Hat LVM Storage yet. In this article we present the current method to use an encrypted device in LVM storage. Future upgrades might introduce new methodology to support this procedure. 

Concepts and components

The following are important concepts and components involved in block device encryption.

Encryption at rest

When talking about data at rest, we refer to data stored in a hard drive. The data we would like to protect is the data stored in Red Hat OpenShift Persistent Volumes, which is neither transferred over the network nor loaded in memory. Having this data encrypted provides protection—if an attacker removes the hard drive from the server and tries to access the data, it won’t be possible without the encryption key. Another advantage of encryption at rest is the fact that a user doesn’t have to do anything to ensure that the data is encrypted, the platform provides an encryption layer that is transparent to the application.


Trusted Platform Module (TPM) is a standard for a security-enhanced cryptoprocessor designed to help improve hardware security using cryptographic keys. In our case, TPM is the chip integrated in the machine where we store the public and private keys used to encrypt a JWE (JSON Web Encryption format) file where the disk encryption key is stored. The key pair stored in the TPM chip is called pin.

This chip can also store Platform Configuration Registers (PCRs), which are hashes of the platform configuration, for example, the UEFI configuration or hardware inventory. Setting the PCRs values allows third party software to create trustworthy computing environments by providing attestation of system state.

In this article we are not going to cover anything related to PCRs, we only use the TPM chip to store the encryption key.

The example in this article is using a TPM2 chip on a physical server. It's also possible to use a virtual TPM (vTPM) to emulate the chip features when working with virtual machines (VMs). One of the available packages for this is swtpm.

Linux Unified Key Setup (LUKS)

LUKS is a disk encryption specification that implements a standard on-disk format to store encryption keys. It was originally intended for Linux, but is also used by different operating systems. LUKS allows performing block device encryption. LUKS2 is currently the default format in Red Hat Enterprise Linux (RHEL) and Red Hat Enterprise Linux (RHEL) CoreOS.

Mounting an encrypted volume requires decrypting it first using a passphrase. An automated process to decrypt a LUKS volume, using a pin stored in a TPM2 chip, is described in the Policy Based Decryption section below.

There are different ways to use LUKS  in conjunction with LVM. In this article we  focus on the use of LVM on LUKS by creating logical volumes on top of an unlocked LUKS container. A kernel subsystem called dm-crypt is used in LUKS to expose encrypted devices as virtual blocks that can be used by LVM.

More information about block device encryption using LUKS in RHEL can be found in the official RHEL documentation.

Policy-Based Decryption (PBD)

Policy-Based Decryption is a collection of technologies to unlock hard drives on physical and virtual machines. Mounting volumes on an encrypted device requires decrypting it first. This process is done automatically in RHEL CoreOS at boot time. As mentioned before, TPM2 is used to store the encryption keys of the JWE file containing the disk encryption key. 

For decryption a pluggable framework called Clevis is used. Clevis handles decryption of LUKS volumes by using the key stored in the JWE file. Clevis also supports the use of other technologies for unlocking volumes, such as tang or sss. More information about policy-based decryption can be found in the official RHEL documentation.

LVM Storage

The main OpenShift storage operator offered by Red Hat is OpenShift Data Foundation, a meta operator that supports other operators such as OpenShift Container Storage to deploy Ceph. Ceph is only supported in a cluster with at least 3 nodes. 

Another available storage solution from Red Hat that can be used in a SNO deployment is the Local Storage Operator (LSO). This operator offers local devices on the node as volumes to the pods, without the implementation of some of the advantages of Container Storage Interface (CSI) such as snapshots, volumes resizing, and so on. 

To fill those gaps and be able to use a powerful storage solution on SNO, the LVM Storage Operator was developed. Saying that, it is possible to use LVM Storage on a Multi Node OpenShift (MNO) deployment, but it is only useful in some specific use cases which are not covered in this article. 

LVM Storage Operator uses LVM to create volume groups and logical volumes, which are exposed as persistent volumes on the pods. LVM Storage Operator eases the management of persistent volumes by leveraging standard Kubernetes CSI features.

Hardware / software diagram

Encryption at the edge

In this article we focus on SNO at the edge, but this particular configuration can also be applied in various other OpenShift setups. Our demo SNO is running vDU workloads of a 5G RAN infrastructure on top of an HPC ProLiant DL110 server, with TPM2 chip, OpenShift version 4.13.0 and LVM Storage Operator 4.13.0.

Storing data in an encrypted device is a common regulation for telecommunications infrastructures. There are multiple techniques and solutions to encrypt data in a disk, the technique described in this article is not the silver bullet to mitigate all the threats on our data, but adds protection against some specific threats. For example, if a disk is replaced by a spare, regulations may require the supplier to destroy that device to avoid data leaks. 

In this case, the fact that the disk is encrypted helps—once the disk is separated from the server, there is no access to the data as the disk encryption key is still in a JWE file encrypted by keys stored in the TPM chip. Having an encrypted disk can also be effective in other physical attacks such as disk theft, as it is easier to steal a single disk than the whole server with TPM. While in this article we are using TPM, there are other mechanisms to store the pin outside the server like tang or sss that are mentioned earlier. Another possibility for enhancing the security of the keys is the use of PCRs with TPM.

If a whole server is stolen, an attacker can decrypt the disk using TPM. For that case encryption alone is not enough, but can be used in conjunction with other techniques such as PCR, which is out of scope for this article.


Encrypted partition configuration

An encrypted partition can be configured at install time by adding the following MachineConfig to the manifest files that create additional configurations. These manifest files are part of the iso image that will be used to install the SNO. More information about manifest configuration files can be found in the official OpenShift documentation.

kind: MachineConfig
  name: 98-encrypted-disk-partition-master
  labels: master
      version: 3.2.0
        - device: /dev/nvme0n1
          wipeTable: false
            - sizeMiB: 204800
              startMiB: 600000
              label: application
              number: 5
        - clevis:
            tpm2: true
          device: /dev/disk/by-partlabel/application
          name: application
          - --cipher
          - aes-cbc-essiv:sha256
          wipeVolume: true

LVM Storage installation

You can install LVM Storage Operator on a 4.13 SNO by applying the following resources using the Operator Lifecycle Manager (OLM) in the OpenShift Container Platform web console or the OpenShift command-line interface (CLI):

$ cat << EOF | oc apply -f -
apiVersion: v1
kind: Namespace
  labels: "true" privileged privileged privileged
  name: openshift-storage
kind: OperatorGroup
  name: openshift-storage-operatorgroup
  namespace: openshift-storage
  - openshift-storage
kind: Subscription
  name: lvms-operator
  namespace: openshift-storage
  installPlanApproval: Automatic
  name: lvms-operator
  source: redhat-operators
  sourceNamespace: openshift-marketplace

Verify that the Cluster Service Version (CSV) is ready:

$ oc get csv -n openshift-storage
lvms-operator.v4.13.0  LVM Storage  4.13.0             Succeeded

More info about LVM Storage on Single Node OpenShift can be found in the official OpenShift documentation.

LVM Storage custom resources

The LVMCluster CR represents the volume groups that should be created and managed on selected devices (if no device paths are specified, all available disks will be used). Through LVMCluster CR the LVM Operator can ensure the required volume groups are available for use by the applications running on OpenShift.

This article uses the previously created encrypted LUKS partition, available in /dev/mapper/application.

Before applying the LVMCluster object, get the status of the block devices by logging on to the node and using the lsblk command:

sh-4.4# lsblk --fs
NAME        FSTYPE  LABEL  UUID                             MOUNTPOINT
|-nvme0n1p2 vfat    EFI-SYSTEM 4C0F-3022                        
|-nvme0n1p3 ext4    boot   16ebb796-07e4-424f-9a6d-31ea9daec5ec /boot
|-nvme0n1p4 crypto_LUKS        6bd0d5b7-67a4-44b4-87a4-e25b2a01c7c0
| `-root    xfs     root   848adc26-5520-4387-8f27-272ad133f654 /sysroot
`-nvme0n1p5 crypto_LUKS        05bd09eb-4dac-47af-88d8-bf8cdbb6b13c

The nvme0n1p5 device is the previously configured LUKS encrypted partition and has an empty virtual block device called application handled by dm-crypt.

Note: The root partition encryption was done during OpenShift installation. More info about boot disk encryption at install time can be found in the official OpenShift documentation.

If the disk was in use before, it might contain the metadata of a previous filesystem, which will make the next step fail. In this case wipe the content of the disk using the following command:

sgdisk --zap-all /dev/mapper/application

Apply the following to configure the LVMCluster object:

$ cat << EOF | oc apply -f -
kind: LVMCluster
  name: my-lvmcluster
  namespace: openshift-storage
    - name: vg1
        - /dev/mapper/application
        name: thin-pool-1
        sizePercent: 90
        overprovisionRatio: 10

Verify that the LVMCluster CR is in a ready state:

$ oc -n openshift-storage get -o jsonpath='{.items[*].status.deviceClassStatuses[*]}' | jq
  "name": "vg1",
  "nodeStatus": [
      "devices": [
      "node": "super1",
      "status": "Ready"

Verify that the pods in the openshift-storage namespace are running:

$ oc -n openshift-storage get pods
NAME                              READY  STATUS    RESTARTS   AGE
lvms-operator-665f584c68-jx8p9    3/3 Running   0          43m
Topolvm-controller-5496f5d4f4-pskzp 5/5 Running   0          84s
topolvm-node-vq4c2                4/4 Running   0          84s
vg-manager-kwfzz                  1/1 Running   0          84s

Ensure that the storage class is created:

$ oc get storageclass
lvms-vg1 (default)  Delete        WaitForFirstConsumer
true                 5m

Next, log in to the node and review the changes done by the operator:

sh-4.4# lsblk --fs
NAME                      FSTYPE  LABEL  UUID                               MOUNTPOINT
|-nvme0n1p2               vfat    EFI-SYSTEM 4C0F-3022                          
|-nvme0n1p3               ext4    boot   16ebb796-07e4-424f-9a6d-31ea9daec5ec   /boot
|-nvme0n1p4               crypto_LUKS        6bd0d5b7-67a4-44b4-87a4-e25b2a01c7c0   
| `-root                  xfs     root   848adc26-5520-4387-8f27-272ad133f654   /sysroot
`-nvme0n1p5               crypto_LUKS        05bd09eb-4dac-47af-88d8-bf8cdbb6b13c   
  `-application           LVM2_member        W8RFWF-ofT9-8oUF-ylR1-18Cc-j0GD-rschAQ
| `-vg1-thin--pool--1                                                               

The LUKS container now has a new entry of type LVM2_member. This is the Volume Group created by the operator.

Look at the Volume Groups and Logical Volumes:

sh-4.4# vgs
  VG  #PV #LV #SN Attr   VSize   VFree
  vg1   1   1   0 wz--n- 199.98g 20.00g
sh-4.4# lvs
  LV          VG  Attr     LSize   Pool Origin Data%  Meta%  Move Log Cpy%Sync Convert
  thin-pool-1 vg1 twi-a-tz-- 179.80g         0.00   10.42

Application deployment

Now create an application and consume storage from LVM logical volumes. For this example, we use a simple hello-sno application under a testing namespace.

First, create the test namespace:

$ oc create namespace test-lvms

Apply the following configuration to create the PersistentVolumeClaim (a storage request) and a Deployment to bring up one hello-sno pod in the testing namespace and use the claim as a volume:

$ cat << EOF | oc apply -f -
apiVersion: v1
kind: PersistentVolumeClaim
  name: test-pvc
  namespace: test-lvms
  - ReadWriteOnce
  volumeMode: Filesystem 
      storage: 100Gi 
  storageClassName: lvms-vg1
apiVersion: apps/v1
kind: Deployment
  name: hello-sno
  namespace: test-lvms
  replicas: 1
      app: test
    type: Recreate
        app: test
      - name: hello-sno
        command: ["sh", "-c"]
        args: ["echo Hello SNO! >> /data/hello-sno.txt && sleep 99999999"]
          - name: data
            mountPath: /data
        - name: data
            claimName: test-pvc

Next, verify the Persistent Volume Claim has been bound correctly:

$ oc -n test-lvms get pvc
NAME     STATUS   VOLUME                                 CAPACITY   ACCESS MODES   STORAGECLASS   AGE
test-pvc Bound    pvc-bd1ca6d3-c5b5-4559-af99-0ed7e869901b  100Gi      RWO          lvms-vg1       5m35s

Check the application pod is running:

$ oc -n test-lvms get pods
NAME                       READY   STATUS    RESTARTS   AGE
hello-sno-847d4dcc-t5swt   1/1     Running   0          3m12s

Review the block devices to see that the Logical Volume has been created correctly within the LUKS container:

sh-4.4# lsblk --fs
NAME                                               FSTYPE  LABEL  UUID                               MOUNTPOINT
|-nvme0n1p2                                        vfat    EFI-SYSTEM 4C0F-3022                          
|-nvme0n1p3                                        ext4    boot   16ebb796-07e4-424f-9a6d-31ea9daec5ec   /boot
|-nvme0n1p4                                        crypto_LUKS        6bd0d5b7-67a4-44b4-87a4-e25b2a01c7c0   
| `-root                                           xfs     root   848adc26-5520-4387-8f27-272ad133f654   /sysroot
`-nvme0n1p5                                        crypto_LUKS        05bd09eb-4dac-47af-88d8-bf8cdbb6b13c   
  `-application                                    LVM2_member        W8RFWF-ofT9-8oUF-ylR1-18Cc-j0GD-rschAQ
| `-vg1-thin--pool--1-tpool                                                                                  
|   |-vg1-thin--pool--1                                                                                      
|   `-vg1-7c7b8882--13d8--4439--86f2--7bc71a51715f                                                           


Encryption is one of the most used techniques to protect application data against unwanted access. Even though the technique does not mitigate all the possible attacks, it is an important component of an overall security strategy.

As described in this article, using the LVM Storage Operator with an encrypted device offers a solid solution to mitigate some of the threats on Telco workloads running at the edge, and to comply with business regulations.

Sobre os autores

Daniel Chavero Gaspar is an open source enthusiast with more than 20 years working in IT. He joined Red Hat in 2021 to the Telco Engineering team. He has been working with Kubernetes/OpenShift during the last few years and was in a DevOps role for machine learning projects in his previous company.

Read full bio

Navegue por canal

automation icon


Últimas novidades em automação de TI para empresas de tecnologia, equipes e ambientes

AI icon

Inteligência artificial

Descubra as atualizações nas plataformas que proporcionam aos clientes executar suas cargas de trabalho de IA em qualquer ambiente

open hybrid cloud icon

Nuvem híbrida aberta

Veja como construímos um futuro mais flexível com a nuvem híbrida

security icon


Veja as últimas novidades sobre como reduzimos riscos em ambientes e tecnologias

edge icon

Edge computing

Saiba quais são as atualizações nas plataformas que simplificam as operações na borda

Infrastructure icon


Saiba o que há de mais recente na plataforma Linux empresarial líder mundial

application development icon


Conheça nossas soluções desenvolvidas para ajudar você a superar os desafios mais complexos de aplicações

Original series icon

Programas originais

Veja as histórias divertidas de criadores e líderes em tecnologia empresarial