The MDM Hub uses the Hub Server and the Process Server for search. The Elasticsearch client is embedded in the Hub Server to perform real-time indexing. The Process Server performs only the initial indexing after the initial data load. Whether you install all the MDM Hub components on a single machine or on multiple machines, you must configure all the Hub Server and Process Server instances for search.
You can install Elasticsearch on any machine where the MDM Hub components are installed or on a separate machine. Elasticsearch requires an exclusive supported version of the Java Virtual Machine (JVM), which the MDM Hub components do not share.
To configure the MDM Hub components to search with Elasticsearch, set up the Elasticsearch cluster. A cluster is a collection of one or more nodes. Each node is a single server that stores your data and participates in indexing and search operations. Based on your MDM Hub topology and the amount of data to index, you can configure one or more nodes for the cluster. Each node can in turn have multiple indexes. To prevent data loss and maintain cluster stability, you can configure Zen Discovery for multi-node clusters. For disaster recovery, run the Initially Index Smart Search Data batch job, which indexes all the values of the searchable fields in a business entity.
The following image shows a sample installation topology configured for search: