Scaling up your Watson Speech services datastores

Important: IBM Cloud Pak® for Data Version 4.8 will reach end of support (EOS) on 31 July, 2025. For more information, see the Discontinuance of service announcement for IBM Cloud Pak for Data Version 4.X.

Upgrade to IBM Software Hub Version 5.1 before IBM Cloud Pak for Data Version 4.8 reaches end of support. For more information, see Upgrading from IBM Cloud Pak for Data Version 4.8 to IBM Software Hub Version 5.1.

You can scale up the number of replicas for each of your datastores as needed to support your workload.

Permissions you need for these tasks:
You must be an administrator of the Red Hat® OpenShift® project to scale up your installation.

Before you begin

Keep the following in mind when scaling up your installation:

  • All deployments and Kubernetes resources are managed by the Speech operator.
  • Do not directly edit any Kubernetes resources (for example, deployments) except for the WatsonSpeech instance.
  • Make all changes to your deployments only by editing the Speech custom resource.
  • Your configuration is physically limited by the amount of hardware resources available in your Kubernetes cluster and namespace.
  • The PostgreSQL and RabbitMQ datastores are installed only with some microservices. For information about the microservices that depend on these datastores, see Datastores properties.

Datastores scaling topics

Use the following procedures to scale up your PostgreSQL and RabbitMQ datastores. The steps involve modifying the custom resource to increase the number of replicas that are run for your datastores.

Note: For instructions on scaling the Multicloud Object Gateway, refer to the RedHat Multicloud Object Gateway documentation

To scale up your datastores, you edit your custom resource. All of the procedures described in the following topics require you to edit your custom resource. For more information, see Editing the custom resource.

Updating the t-shirt size for your installation as described in Scaling up your installation has no effect on the number of replicas that are used by your datastores. For maximum flexibility, scaling up the Speech services is completely separate from increasing the number of replicas for your datastores.

Scaling up the PostgreSQL datastore

By default, EDB PostgreSQL is installed with three replicas for high availability. Each replica is typically scheduled within a different Kubernetes worker node if resources allow. For information about the data that PostgreSQL stores, see Configuring the PostgreSQL datastore.

You scale up the PostgreSQL datastore on an already running solution by changing the number of replicas in the Speech custom resource:

  1. Look up the resource name for the Speech service. The command returns the name of the WatsonSpeech resource. 
    kubectl get WatsonSpeech
  2. Edit the object by specifying the name that is returned in the previous step: 
    kubectl edit WatsonSpeech name
  3. Add or edit, if already exists, the global.datastores.postgresql.replicas field to the desired number of replicas.
  4. Save and close the file in the editor to apply your configuration changes.
  5. The speech operator syncs your changes with the Kubernetes cluster. It may take from 30 minutes to 1 hour for your cluster to reflect the changes.

Scaling up the RabbitMQ datastore

By default, RabbitMQ is installed with three replicas for high availability. Each replica is typically scheduled within a different Kubernetes worker node if resources allow. Before performing the installation, you can configure the number of replicas and the CPU and memory resources for each replica via the Speech custom resource. For information about the data that RabbitMQ stores, see Configuring the RabbitMQ datastore.

You scale up the RabbitMQ datastore on an already running environment by changing the number of replicas in the Speech custom resource:

  1. Look up the resource name for the Speech service. The command returns the name of the WatsonSpeech resource. 
    kubectl get WatsonSpeech
  2. Edit the object by specifying the name that is returned in the previous step: 
    kubectl edit WatsonSpeech name
  3. Add or edit, if already exists, the global.datastores.rabbitmq.replicas field to the desired number of replicas.
  4. Save and close the file in the editor to apply your configuration changes.
  5. The speech operator syncs your changes with the Kubernetes cluster. It may take from 30 minutes to 1 hour for your cluster to reflect the changes.