Pipeline nodes are the physical machines that host one or more
pipeline processes.
The pipeline node is where you install and start the pipeline executable
that runs the pipeline processes. You configure and maintain the pipeline
configuration file for all pipelines that are hosted by this machine. The
system also writes the pipeline messages to the log files on the pipeline
nodes.
Pipeline nodes connect pipeline processes to these components
of the product architecture:
- Acquisition programs
- As part of the extract, transform, and load (ETL) process, acquisition
programs use transports to send UMF data into pipelines for processing. You
use the transport method appropriate to the type of acquisition program to
connect to the pipelines. For example, if you use the UMF file utility as
an acquisition program, you use the file transport.
- Entity database
- The entity database contains entity information. Pipelines access entity
information while processing incoming records for entity and relationship
resolution. The pipeline node must have the appropriate database client installed
and configured, so that the pipelines can access the entity database.
- Queues
- If your system uses queues as transport methods to send data to the pipelines
for processing, you must install and configure the appropriate message queuing
software on each pipeline node.
- For example, if your pipelines run on Microsoft® Windows®, you must install Microsoft
MQ on each pipeline node; otherwise, the pipelines on this pipeline node will
not run.
- Application monitor and the application monitor database
- If configured, you can use the application monitor to get status and statistics
for registered pipelines that are running on each pipeline node. You can also
use the application monitor to route the processed data (such as alerts) to
other databases or systems. Routing requires that the pipeline be able to
connect to the application monitor component of the Configuration Console
through an HTTP transport for the routing instructions.
- If you use the application monitor to monitor pipeline status, you must
install and configure an SNMP agent on each pipeline node to send status and
statistics about each registered pipeline to the Configuration Console.
- Address hygiene servers
- If your system uses address hygiene products from other companies for
additional address cleansing, each pipeline node must be configured to connect
to the address hygiene servers.
- Web services
- If your system is configured to use Web services, you must use an HTTP
transport to connect the pipeline processes on the pipeline node to the Web
services.