Data Prepper Introduction
Data Prepper is a server side data collector capable of filtering, enriching, transforming, normalizing and aggregating data for downstream analytics and visualization.
Data Prepper lets users build custom pipelines to improve the operational view of applications. Two common uses for Data Prepper are trace and log analytics. Trace analytics can help you visualize the flow of events and identify performance problems, and log analytics can improve searching, analyzing and provide insights into your application.
Data Prepper is compromised of one or more pipelines that collect and filter data based on the components set within the pipeline. Each component is pluggable, enabling you to use your own custom implementation of each component. These components include the following:
- One source
- One or more sinks
- (Optional) One buffer
- (Optional) One or more processors
A single instance of Data Prepper can have one or more pipelines.
Each pipeline definition contains two required components: a source and sink. If buffers and processors are missing from the Data Prepper pipeline, Data Prepper uses the default buffer and a no-op processor.
The source is the input component that defines the mechanism through which a Data Prepper pipeline will consume events. A pipeline can have only one source. The source can consume events either by receiving the events over HTTP or HTTPS or by reading from external endpoints like OTeL Collector for traces and metrics and Amazon Simple Storage Service (Amazon S3). Sources have their own configuration options based on the format of the events (such as string, JSON, Amazon CloudWatch logs, or open telemetry trace). The source component consumes events and writes them to the buffer component.
The buffer component acts as the layer between the source and the sink. A buffer can be either in-memory or disk based. The default buffer uses an in-memory queue called
bounded_blocking that is bounded by the number of events. If the buffer component is not explicitly mentioned in the pipeline configuration, Data Prepper uses the default
The sink is the output component that defines the destination(s) to which a Data Prepper pipeline publishes events. A sink destination could be a service, such as OpenSearch or Amazon S3, or another Data Prepper pipeline. When using another Data Prepper pipeline as the sink, you can chain multiple pipelines together based on the needs of the data. A sink contains its own configuration options based on the destination type.
Processors are units within the Data Prepper pipeline that can filter, transform, and enrich events using your desired format before publishing the record to the sink component. The processor is not defined in the pipeline configuration; the events publish in the format defined in the source component. You can have more than one processor within a pipeline. When using multiple processors, the processors are run in the order they are defined inside the pipeline specification.
Sample pipeline configurations
To understand how all pipeline components function within a Data Prepper configuration, see the following examples. Each pipeline configuration uses a
yaml file format.
This pipeline configuration reads from the file source and writes to another file in the same path. It uses the default options for the buffer and processor.
sample-pipeline: source: file: path: <path/to/input-file> sink: - file: path: <path/to/output-file>
The following pipeline uses a source that reads string events from the
input-file. The source then pushes the data to the buffer, bounded by a max size of
1024. The pipeline is configured to have
4 workers, each of them reading a maximum of
256 events from the buffer for every
100 milliseconds. Each worker runs the
string_converter processor and writes the output of the processor to the
sample-pipeline: workers: 4 #Number of workers delay: 100 # in milliseconds, how often the workers should run source: file: path: <path/to/input-file> buffer: bounded_blocking: buffer_size: 1024 # max number of events the buffer will accept batch_size: 256 # max number of events the buffer will drain for each read processor: - string_converter: upper_case: true sink: - file: path: <path/to/output-file>