Integrating with IBM Business Automation Insights

New in 8.10.1 IBM® Business Automation Insights is a platform-level component of the IBM Digital Business Automation for Multicloud platform that gives business owners and data analysts insights into their business data.

When Operational Decision Manager integrates with Business Automation Insights, the platform captures all decisions that are generated by the execution of rulesets as events. You can aggregate these emitted events into business-relevant KPIs and present them in meaningful dashboards for lines of business to get a real-time view on their business operations. You can also store the data in your data lake, and use data analytics tools to gain further insights.

For Business Automation Insights to capture and process Operational Decision Manager events, you must configure each side of the integration.

  1. Set up Business Automation Insights and note the Kafka server details, which are needed when you enable the ODM event emitter.

    For more information, see IBM Business Automation Insights Kubernetes deployment specifics.

  2. Configure Decision Server Rules to emit events to Business Automation Insights.

    For Decision Server Rules, you enable the ODM event emitter in the execution unit (XU) deployment descriptor or by using the res-setup Ant task. You can also create a plugin-configuration.properties file to store and share all of the configuration parameters that are needed to connect to the Kafka cluster. For more information, see Enabling the ODM event emitter.

    For ODM on Certified Kubernetes, you configure ODM for production with a Kubernetes secret that contains the plugin-configuration.properties file. For more information, see Configuring the ODM event emitter.

  3. Add the appropriate ruleset properties to each ruleset for which you want to emit events.

    For more information, see Built-in ruleset properties for the ODM event emitter.

  4. Visualize your business data that is stored in Elasticsearch in the form of Kibana dashboards.

    For more information, see Decisions dashboard.

The ODM event emitter supports the following features of Operational Decision Manager:

  • Decision engine
  • Decision service
  • Decision model

The ODM event emitter has the following limitations:

  • The execution object model (XOM) must be Jackson-compatible. For more information, see JSON serialization of ruleset XOMs.
  • You cannot enable the ODM event emitter and send events to Business Automation Insights in addition to storing data in Decision Warehouse.
  • Decision model services from a previous release can emit events to Business Automation Insights only if you rename the main flow task of the decision model to DCDMMainFlowTask in the Business console. A compilation is done as a result of the rename. You then need to set the bai.emitter.enabled ruleset property to true to enable the ruleset to emit events.
  • The storage of ODM time series in Elasticsearch is limited by the number of fields in type mappings. The performance degrades as the number of fields increases. By default, Elasticsearch is limited to 1000 fields. For more information, see https://www.elastic.co/guide/en/elasticsearch/reference/master/mapping.html#mapping-limit-settings. When this limit is reached, Elasticsearch raises the following exception: "java.lang.IllegalArgumentException: Limit of total fields [1000] in index [odm-timeseries-write] has been exceeded". If you include ruleset input or output parameters in the ODM event emitter configuration, you must monitor the total number of indexed fields to avoid reaching the limit, see https://www.elastic.co/guide/en/elasticsearch/reference/current/indices-get-mapping.html.
  • The throughput of ruleset execution might be slower when event emission is enabled on a ruleset. The overhead depends on various factors: configuration of the ODM event emitter, ruleset characteristics (number of input/output parameters, number of executed rules and tasks, if the emitter configuration includes these elements in the emitted payload), network bandwidth between the ODM event emitter and Kafka. There should be enough memory available for the event emitter, in particular if it is configured to use Kafka producer buffers. A properly-sized, properly-monitored, high-performance, and highly-available Kafka cluster is a key factor for reducing the overhead to a minimum.
  • Due to Elasticsearch field mapping requirements, two versions of the same ruleset in a given rule application cannot use input or output parameters with the same name but different types. Input or output parameters of different rulesets, or in different rule applications, can use the same name but different types.