Scaling IBM MDM Publisher
IBM® MDM Publisher supports vertical or horizontal scaling to increase performance and allow it to load larger data sets.
Horizontal scaling
MDM Publisher
supports horizontal scaling by using a scalable Spark pod with a single
mdm-publisher
container.
To scale the MDM Publisher bulk load operation up or down, depending on the dataset size you are trying to process, modify the number of executors that MDM Publisher will spawn, or modify their corresponding CPU and memory parameters. For more information about changing the number of executors, see the publish-config.xml property descriptions in Configuring IBM MDM Publisher.
Vertical scaling
MDM Publisher also supports vertical scaling. To increase MDM Publisher performance and enable loading of larger data sets, you can add more resources to the Kubernetes nodes and increase the resource allotment for your MDM Publisher deployment.
- For Red Hat OpenShift deployment configuration: ${INSTALL_LOC}/mdm-publisher/ibm-publisher-services-prod/values-openshift.yaml
- For Kubernetes deployment configuration: ${INSTALL_LOC}/mdm-publisher/ibm-publisher-services-prod/values-k8s.yaml
- For Minikube deployment configuration: ${INSTALL_LOC}/mdm-publisher/ibm-publisher-services-prod/values-minikube.yaml
- Edit the YAML file that corresponds to your deployment type.
- Modify the following configuration
section:
publisher: image: mdm-publisher imageTag: 0.1.623 productTag: 0.1.623 imagePullPolicy: IfNotPresent deployment: resources: requests: cpu: 4 memory: 8Gi limits: cpu: 4 memory: 8Gi
- After completing your changes and saving the file, run the following command to apply the
changes to your MDM Publisher
instance:
${INSTALL_LOC}/mdm-publisher/bin/update_configuration.sh
- When prompted with the message As part of this MDM Publisher configuration update, are you also updating endpoints? (Yy/Nn), reply N. This ensures that only the properties files are updated.
- Wait for the
mdm-publisher-0
pod to re-enter aREADY
state before trying to access it again.