WebSphere clustering guidelines

Preparing the data sources

In addition to any other instructions provided in this guide, complete the following tasks for your data sources:

• The Unica Campaign database must be on a server that is accessible to all servers in the cluster, but it does not need to be on a server in the cluster.
• When you configure the JDBC provider, specify the cluster as the scope.

Installing the products

When you run the installer, be sure that you install Unica Platform and Unica Campaign only once, on a server that is designated as the administration server for the cluster. The administration server is accessible to all servers in the Unica Campaign cluster.

You do not need to install the software on each cluster member. Instead, you install it on the administration server, create the EAR or WAR file, and then deploy the EAR or WAR file on each cluster member.

Pre-deployment steps

Before you deploy Unica Campaign, complete the following task:

• Share the top-level directory in which Unica Campaign is installed. For example, if you installed Unica Campaign in C:\CampaignCluster\UNICA\Campaign, you would share the entire CampaignCluster directory.

Deployment steps

In addition to the instructions in the deployment chapter (Deploying the Unica Campaign web application), complete the following tasks:

1. Map modules to servers. When you set the options in the Select Installation Options wizard for WebSphere®, select the cluster and the web server when mapping modules to servers.
2. Extra instructions for the Generic JVM property: Configure the Generic JVM property on each server in the cluster.

   The path that you specify in the <CAMPAIGN_HOME> and other properties must point to the shared installation directory.

   Set the following extra parameters for clusters:

   • Set the following parameter to ensure that Deliver ETL is not triggered by all clustered nodes:

     -Dcampaign.deliveretl.disabled=true

     For the one node that should trigger ETL Unica Deliver data for all other nodes, set the parameter to false.

   • Set the following parameter to ensure that Unica Interact ETL is not triggered by all clustered nodes:

     -Dcampaign.interactetl.disabled=true

     For the one node that should trigger ETL Interact data for all other nodes, set the parameter to false.

   Property to enable/disable Journey CH RH in campaign.

   Property name: campaign.journeyetl.disabled

   Supported values: true or false

   Default value: false

   • -Dcampaign.journeyetl.disabled=false

     If system property is not present or is set to false, campaign will read CH RH messages from Kafka and dump them in CH RH tables.

   • -Dcampaign.journeyetl.disabled=true

     If system property is set to true, campaign will not read the CH RH data from Kafka and no data will be dumped in Campaign Database.

Post-deployment steps

If you are using a plug-in for load balancing, complete the following configuration steps:

• For Unica Campaign to work effectively in a clustered environment, users must remain on a single node for their entire session. This option for session management and load balancing is known as session affinity. The documentation for your application server provides details about configuring your installation to use session affinity.
  Note: When a node fails on a system that uses this configuration option, every user session on that node also fails. The load balancer must not switch users to another available node because user authentication applies only to a single node in Unica Campaign. Users are prompted to log in again. In some cases, an unexpected error and corresponding data loss can occur.
• Log in to Unica Campaign. Select Settings > Configuration, and configure the following URL parameters so that all references to the Unica Campaign server use the proxy host and port:

  Campaign | navigation | serverURL