Each Amazon Kinesis connector application is a pipeline that determines how records from an Amazon Kinesis stream will be handled. Records are retrieved from the stream, transformed according to a user-defined data model, buffered for batch processing, and then emitted to the appropriate AWS service.
A connector pipeline uses the following interfaces:
* __Pipeline:__ The pipeline implementation itself.
* __Transformer:__ Defines the transformation of records from the Amazon Kinesis stream in order to suit the user-defined data model. Includes methods for custom serializer/deserializers.
* __Filter:__ Defines a method for excluding irrelevant records from the processing.
* __Buffer:__ Defines a system for batching the set of records to be processed. The application can specify three thresholds: number of records, total byte count, and time. When one of these thresholds is crossed, the buffer is flushed and the data is emitted to the destination.
* __Emitter:__ Defines a method that makes client calls to other AWS services and persists the records stored in the buffer. The records can also be sent to another Amazon Kinesis stream.
Default logging is handled by [go-kit package log](https://github.com/go-kit/kit/tree/master/log). Applications can override the default loging behaviour by implementing the [Logger interface][log_interface].