12.11. Examples

12.11.1. Ingest a GeoMesa Kafka Topic into HBase

A common use case is to store the most recent state of an object in a GeoMesa Kafka data store to drive a live map view, and to persist every state of the object into long-term storage in HBase or Accumulo for history and analytics. This example will show how to pull data off a GeoMesa Kafka topic, and persist it in HBase with minimal configuration.

12.11.1.1. Install the GeoMesa NARs

Follow the steps under Installation to install the GeoMesa processors in your NiFi instance. This tutorial requires the GeoMesa Kafka and HBase NARs, as well as the standard GeoMesa service NARs.

12.11.1.2. Add and Configure the Processors

The first step is to add two processors to your flow, a GetGeoMesaKafkaRecord processor and an AvroToPutGeoMesa processor. Connect the output of the Get processor to the input of the Put processor:

For a robust system, you would want to add further processing for handling successes and failures. For this example, just auto-terminate all other connections for simplicity.

Next, create a new controller service by clicking ‘Configure’ on the NiFi flow, then going to the Controller Services tab and clicking the + button, then selecting the GeoAvroRecordSetWriterFactory. The GeoAvroRecordSetWriterFactory can be left with its default configuration, which is to use NiFi expressions. The GetGeoMesaKafkaRecord processor will populate the necessary attributes according to the feature type being read:

The GetGeoMesaKafkaRecord processor will pull data off of the GeoMesa Kafka topic and write it out using a NiFi record writer. The processor can output data in any format supported by the NiFi records API, but for this example we’re going to use GeoAvro to simplify the ingest into HBase. Configure the processor by setting the appropriate Kafka connection parameters and setting the GeoAvroRecordWriterFactory you just created for the output. Set a unique Kafka group ID, to ensure that the processor reads all the data coming from the topic:

The AvroToPutGeoMesa processor needs to be configured with a DataStoreService for HBase. Go back to the Controller Services tab in NiFi, and add another service, this time selecting HBaseDataStoreService. You will need to configure it with the path to the hbase-site.xml file for your cluster, which will let it connect to HBase, and a catalog table, which is where it will write out data:

Once the controller service is enabled, go back to configure the AvroToPutGeoMesa processor with the HBaseDataStoreService. Set Use provided feature ID to false, as we want to keep all historical features, instead of just the latest instance of each feature:

Once all the processors and controller services are configured, enable them in the NiFi UI. If everything goes correctly, you’ll start seeing your Kafka data show up in HBase.