Installing Fusion Data Foundation

Install Fusion Data Foundation in local, dynamic, external, client, or provider modes.

Before you begin

  • Important: If you install Fusion Data Foundation 4.21 or later and the Global Data Platform remote mount service is installed and configured, the service is automatically migrated to version 6.0.1.0. Review the supported migration scenarios before installation.
  • Existing storage installations

    If you already have an installation of Fusion Data Foundation cluster or Red Hat® OpenShift® Data Foundation cluster, the IBM Fusion UI automatically discovers it.

  • Storage device and node requirements
    • Use a maximum of nine storage devices per node to ensure optimal recovery time and stay within cloud provider dynamic storage device attachment limits. Using more storage devices increases recovery time if a node is lost.
    • Add nodes in multiples of three, with each node in a different failure domain.
    • For deployments with three failure domains, scale up storage by adding disks in multiples of three, with the same number of disks from nodes in each failure domain.
  • Clean installation verification

    Before enabling the Fusion Data Foundation service, verify that there is no previously installed Fusion Data Foundation, Red Hat OpenShift Data Foundation, OpenShift Container Storage, or Rook-Ceph in this OpenShift cluster:

    1. Verify that the namespace openshift-storage does not exist.
    2. Run the following command to confirm that no storage resources exist: 
      oc get storagecluster -A; oc get cephcluster -A
    3. Run the following command to confirm that no nodes have storage labels: 
      oc get node -l cluster.ocs.openshift.io/openshift-storage=
  • Infrastructure nodes
    Ensure that the following prerequisites are met:
    • For infrastructure requirements, see Infrastructure requirements.
    • For Fusion Data Foundation resource requirements, see ODF sizer tool. For common Fusion Data Foundation components per instance requirements of CPU and Memory, see Resource requirements.
    • If you want to deploy Fusion Data Foundation in an Infra node, make sure the worker node label exists.
      oc get node -l "node-role.kubernetes.io/worker="
    • Ensure that there are no taints in the infra or compute nodes that are used as storage nodes.

    For more information about the limitations and considerations of using OVE as an alternative to OpenShift Container Platform platform, see OVE considerations.

About this task

  • Both Data Foundation service and Global Data Platform service can coexist during the following scenarios:

    • Deploy Data Foundation service without dedicated mode.
    • Deploy Data Foundation in dedicated mode first, and then deploy Global Data Platform service.
  • For Red Hat OpenShift Data Foundation deployed outside of IBM Fusion, you cannot add nodes or capacity from IBM Fusion. See Scaling storage in Red Hat OpenShift Data Foundation documentation.
  • For client cluster with Fusion Data Foundation, you do not install IBM Fusion spoke cluster or Fusion Data Foundation from the user interface. These installations are automatically run from the IBM Fusion HCI hub cluster.

  • Data Foundation service supports five storage types: local, dynamic, external, client, and provider. Only one device type can be used for Data Foundation service in an OpenShift Container Platform. For more information about the different deployment modes, see Storage cluster deployment approaches.

    The following table shows supported platforms and which device types are supported in each platform.
    Platform Support
    VMware Local, dynamic, client, and external device
    Bare Metal Local, external device, client, provider mode
    Linux on IBM Z Local, dynamic, client, and external device
    IBM Power Systems Local, client, and external device
    ROKS on IBM Cloud (ROKS on VPC and Classic Bare Metal) Cloud-managed Fusion Data Foundation service. For the procedure to install, see Understanding Fusion Data Foundation.
    Self managed OpenShift Container Platform on Microsoft Azure Dynamic and client
    Self managed OpenShift Container Platform on Amazon Web Services Dynamic and client
    Amazon Web Services ROSA and Amazon Web Services ROSA HCP Dynamic and client
    Self managed OpenShift Container Platform on Google Cloud Dynamic and client
  • For IBM Fusion services and platform support matrix, see IBM Fusion services and platform support matrix.

Procedure

  1. Go to Services page in IBM Fusion web console.
  2. In the Available section, click the Data Foundation tile.
  3. In the Data Foundation page, go through the details about the service and click Install.
  4. In the Install service window, select the Device type. The available options can be Local, Dynamic, or External based on which option is supported on your platform.

    If you want to use storage through a provider and work with Data Foundation service storage as a client, then there is no Install button available. A notification is displayed asking you to connect to the provider cluster.

  5. Choose whether to enable Automatic updates and click Install.
    Note: If you select Automatic updates, then IBM Fusion auto upgrades Fusion Data Foundation version.

    For example, 4.14.x to 4.14.y, 4.16.x to 4.16.y, and across the subscription channels like 4.14.x to 4.15.y, 4.15.x to 4.16.x. For storage support matrix, see Support matrix.

    However, it does not auto convert Red Hat OpenShift Data Foundation to Fusion Data Foundation.

    The Fusion Data Foundation operator starts to deploy and you can see Data Foundation in the Installed section of the Services page. Initially, the status shows as Installing and progress of the installation is mentioned in percentage. After successful completion of the installation, the status changes to Running. A Data Foundation page is added to the menu.

  6. To validate the installation, do the following steps in OpenShift Container Platform web console:
    • Go to Installed Operators to check whether the Fusion Data Foundation operator is listed and Status shows as Succeeded.
    • Additionally, if the installed platform is Bare Metal, Linux on IBM Z, or VMware using local device, check whether the Local Storage operators is installed and its status shows as Succeeded.
  7. Verify the Fusion Data Foundation service installation.
    1. To verify the installation of Fusion Data Foundation on IBM Fusion, perform the checks specified in Verify the Fusion Data Foundation.
    2. Additionally, run the command to verify that Fusion Data Foundation service is installed successfully: 
      oc describe odfmanager/odfmanager -n <Fusionns>

      Update <Fusionns> with your namespace value.

    Table 1. Install states Fusion Data Foundation service
    Install State Description
    Installing Fusion Data Foundation installation is ongoing, and there are no errors yet.
    Failing There is an installation error but IBM Fusion retries to install Fusion Data Foundation. If the problem is solved, this status changes to Installing or Completed.
    Completed Fusion Data Foundation installation completed successfully.
    Table 2. Upgrade states Fusion Data Foundation service
    Upgrade State Description
    Not started Fusion Data Foundation upgrade has not started yet.
    Upgrading Fusion Data Foundation upgrade is ongoing, and there are no errors yet.
    Failing There is an installation error but IBM Fusion retries to upgrade Fusion Data Foundation. If the problem is solved, this status changes to Upgrading or Completed.
    Completed Fusion Data Foundation upgrade completed successfully.
    Table 3. Health states Fusion Data Foundation service
    State Description
    Installing Fusion Data Foundation installation is ongoing. For more information about installation status details, see Table 1.
    Upgrading Fusion Data Foundation upgrade is ongoing. For more information about upgrading status details, see Table 2.
    Healthy Fusion Data Foundation installation completed successfully and the service is in normal state.
    Degraded Fusion Data Foundation installation completed successfully but the service is not in normal state.

What to do next

Configure the Fusion Data Foundation storage service by following the instructions mentioned in Configuring Fusion Data Foundation.

Optionally, you can also setup encryption to use an external Key Management System. For the procedure to set up encryption, see Preparing to connect to an external KMS server in Fusion Data Foundation.

Note: You can view the Usable capacity and Health for an externally discovered Fusion Data Foundation but cannot configure storage.