Moving to production

When you are satisfied with your test environment and are ready to move to production, some security and scalability settings are necessary.

Before you begin

Before you move to a production work environment, consider the following recommendations.
  1. Read all security instructions in Securing communications and data.
  2. Test the results of your anonymization rules on real events from your process to check whether anonymization works as expected in the data lake. For more information, see Anonymization.
  3. Read the recovery instructions in Preparing your cluster for fault tolerance to make sure that your IBM Business Automation Insights system is set up is for recovery if crashes occur.
  4. Replace the default credentials with your own by creating secrets for the bai-main chart and ibm-dba-ek subchart. For instructions, see Configuring custom secrets.

About this task

For heavy-load use cases, also set up your test environment for a workload similar to what you expect in your production environment. This will help you correctly size your IBM Business Automation Insights deployment according to the number of events that will be ingested and processed in production.

Procedure

To size your test and production environments:

  1. Size your Apache Kafka or IBM Event Streams cluster so that it can ingest all events that IBM Business Automation Insights emitters send. For more information, see your Apache Kafka distribution documentation.
  2. Size the storage space needed for Elasticsearch so that it can handle all the data that IBM Business Automation Insights produces.

    If you use embedded Elasticsearch and Kibana, rather than an external Elasticsearch installation, consider increasing the storage capacity of the respective persistent volumes. See Setting up Elasticsearch storage.

  3. Size the storage space for Apache Flink.

    The storage space needed is proportional to the sum of concurrently open artifacts or assets originating from products of the IBM Cloud Pak for Automation platform, such as BPM processes and activities, cases, decisions, and more.

    Tip: To size persistent volumes for production, you can start from the storage consumption in your test environment.

    Also consider increasing the storage capacity of the persistent volumes and remember to take local disk space for temporary files into account. For more information, see Setting up Apache Flink storage.

  4. Size your Apache Flink and Elasticsearch cluster as instructed in the following iterative process.
    1. Check whether some back pressure is exerted from Elasticsearch to Apache Flink.
    2. If you observe some Elasticsearch back pressure, scale Elasticsearch as explained in Monitoring job throughput and back pressure.
    3. Look up how many events are ingested by the Flink jobs and compare that number to how many events are sent to Apache Kafka.
    4. If IBM Business Automation Insights consumes fewer events than are sent to Kafka, scale IBM Business Automation Insights as explained in Scaling an IBM Business Automation Insights system.
    5. Repeat the process from step 4.a.

What to do next

When you operate your cluster, you might want to regularly monitor whether your system is still correctly configured and sized for your current usage.

In a production environment, it is not a safe practice to use self-signed certificates. Rather, use the HTTPS protocol with trusted certificates. For more information, see Creating an Ingress resource to use your own TLS certificates.