Creating data planes to organize remote physical locations

A data plane is a logical grouping of one or more physical locations. Users can deploy workloads to a data plane. The workload will be scheduled on one of the physical locations associated with the data plane.

Users cannot schedule workloads directly on a remote physical location. The remote physical location must be added to a data plane and the remote physical location must be enabled so that the scheduling service can schedule workloads on the physical location.

Who needs to complete this task?

To complete this task, you must have the Manage data planes permission in IBM® Software Hub.

When do you need to complete this task?

Complete this task if you want to enable users to run workloads on remote physical locations.

About this task

Data planes are flexible. You can create data planes to group remote physical locations in any way that makes sense for your needs.

The following diagram shows examples of how you can use data planes to organize remote physical locations into logical groups.

The architectural diagram is described in the following text.

In this example, the primary instance of IBM Software Hub (the hub) has three data planes.

Data plane 1
This data plane is for workloads that require GPUs.

The data plane contains four remote physical locations. Each remote physical location has between 2 and 4 GPUs. Each remote physical location is located in a different country.

Data plane 2
This data plane is for workloads that must be processed in the European Union.

The data plane contains two remote physical locations. Both physical locations are in countries that are part of the European Union.

Data plane 3
This data plane is for workloads that exceed the available vCPU and memory on the hub.

The data plane contains two remote physical locations. One of the remote physical locations has lower priority, so the scheduling service will schedule workloads on that remote physical location only if it cannot schedule the workload on the remote physical location with higher priority.

You can optionally add the same remote physical location to multiple data planes.

You can create data planes before you have any remote physical locations; however, you cannot enable the data plane, which means that users cannot run workloads on the physical location.

Procedure

  1. Log in to the web client.
  2. From the navigation menu, select Administration > Configurations and settings.
  3. Click Data planes, then click New data plane.
  4. Specify the following details:
    1. Enter a unique identifier for the data plane.
      The ID must meet the following requirements:
      • Be between 1 - 50 characters
      • Contain only lowercase alphanumeric characters, dash (-), or period (.)
      • Start with an alphanumeric character
      • End with an alphanumeric character
    2. Enter a display name for the data plane.
    3. Enter a description for the data plane.
      The description should help users understand when or why they should run workloads on the data plane.
  5. Click Next.
  6. Optional: Add remote physical locations to the data plane.
    You can add only remote physical locations that are already registered with this instance of IBM Software Hub.
    Important: You cannot remove data planes from the data plane after you click Create.
  7. Click Next.
  8. If you added remote physical locations to the data plane, decide whether you want to enable the data plane.

    If you enable the data plane, users can run workloads on the physical locations that are associated with the data plane.

    If you did not add any remote physical locations to the data plane, you cannot enable it.

  9. Click Save.

Results

  • If you enabled the data plane, users run workloads on it.

    You can create service instances on remote physical locations for the following services:

    • Analytics Engine powered by Apache Spark
      Restriction: If you want to create Analytics Engine powered by Apache Spark instances on remote physical locations, you cannot use job interface in Analytics Engine powered by Apache Spark. You must set sparkAdvEnabled to false. For more information about this setting, see Specifying additional configurations for Analytics Engine powered by Apache Spark.
    • DataStage
    • Orchestration Pipelines

      Tech preview This feature is supported as a technical preview.

  • If you did not enable the data plane, users can see the data plane from the Data planes page, but cannot run any workloads on it.