| Prompt: |
| What are the metrics in Apache Beam? |
| |
| Response: |
| In the Apache Beam model, [metrics](https://beam.apache.org/documentation/programming-guide/#metrics) provide insights into the current state of your pipeline, including during pipeline execution. |
| |
| Metrics are named and scoped to a specific step in the pipeline. They can be created dynamically during pipeline execution. If a runner doesn't support some part of reporting metrics, the fallback behavior is to drop the metric updates rather than fail the pipeline. |
| |
| Beam provides a number of [built-in metric types](https://beam.apache.org/documentation/programming-guide/#types-of-metrics): |
| * Counters |
| * Distributions |
| * Gauges |
| |
| To declare a metric, use the `beam.metrics.Metrics` class. For example: |
| |
| ```python |
| self.words_counter = Metrics.counter(self.__class__, 'words') |
| self.word_lengths_counter = Metrics.counter(self.__class__, 'word_lengths') |
| self.word_lengths_dist = Metrics.distribution(self.__class__, 'word_len_dist') |
| self.empty_line_counter = Metrics.counter(self.__class__, 'empty_lines') |
| ``` |
| |
| For implementation details, see the [WordCount example with metrics](https://github.com/apache/beam/blob/master/sdks/python/apache_beam/examples/wordcount_with_metrics.py). |
| |
| You can export metrics to external sinks. Spark and Flink runners support REST HTTP and Graphite. |
