Administering Watson Machine Learning

You can manage the tools and services that are available with the Watson Machine Learning service.

Scaling Watson Machine Learning based on workload

IBM Cloud Pak for Data supports scaling of add-on services. If you are a cluster administrator, you can scale Watson Machine Learning after installation from small (the default configuration) to medium to manage a larger workload.

Scaling changes the capacity of services by adjusting the number of pods that are available. Pods act as servers with a set resource limitation that run an application or function. Pods consume resources such as core and memory when components distribute tasks to them. Scaling the pods to medium, for example, increases the processing capacity of the application.

For details, refer to Scaling services

Configure Watson Machine Learning Accelerator for Deep Learning Experiments

Deep Learning Experiments accelerate the iterative process of training a deep learning model by simplifying the process to train models in parallel with an on-demand GPU compute cluster. To build, train, and deploy deep learning experiments, you must configure Watson Machine Learning Accelerator to work with Watson Machine Learning. For details, refer to Connecting Watson Machine Learning Accelerator to Watson Machine Learning.

Troubleshooting

Known Issue: If you install Watson Machine Learning using a custom CR name other than the default wml-cr, then quiesce/unquiesce and backup/restore commands will fail with this error message:

wmlbases.wml.cpd.ibm.com "wml-cr" not found

Versions Affected: IBM Cloud Pak for Data 4.0.1 WML Assembly

Workaround: Follow this procedure to modify the config maps and then rerun quiesce/unquiesce or backup/restore:

  1. Determine the CR Name in your installation.

     # oc get wmlbases.wml.cpd.ibm.com

 NAME     AGE
 wml-cr   10d

  2. Run oc edit cm cpd-wml-maint-aux-br-cm and replace all occurrences of wml-cr with the CR Name from Step 1.

  3. Run oc edit cm cpd-wml-maint-aux-qu-cm and replace all occurrences of name: wml-cr with the CR Name from Step 1.

Parent topic: IBM Watson Machine Learning