New component: Cloudwatch metric streams encoding extension

[constanca-m]

The purpose and use-cases of the new component

Some aws receivers and exporters, like awsfirehosereceiver, get data from cloudwatch metric streams. The purpose of this extension is to transform these metrics into otel metrics.

Relates to #37113.

Example configuration for the component

extensions:
  cloudwatch_metric_streams_encoding:
     # See output format: https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/aws-resource-cloudwatch-metricstream.html#cfn-cloudwatch-metricstream-outputformat
     # format: opentelemetry1.0 # options [json, opentelemetry1.0], and opentelemetry0.7 not supported.

receivers:
  awsfirehose:
    include_metadata: true
    access_key: "access_key"
    endpoint: :4434
    encoding: cloudwatch_metric_streams_encoding

Telemetry data types supported

• Metrics

Code Owner(s)

@axw

Sponsor (optional)

@atoulme

Additional context

No response

[axw]

Adding to the motivation slightly: the CloudWatch formats (logs & metrics) understood by the awsirehosereceiver could also be used in other receivers, e.g. the S3 receiver, which already supports the encoding extension.

@constanca-m and I discussed offline: my understanding is that since she is not (yet) a member of the OpenTelemetry org, she cannot be a code owner at this time. I'm happy to take on code ownership, and later on I would ideally add @constanca-m as a code owner too.

[Aneurysm9]

As cloudwatch describes a lot of things I think that any encoding extensions that process data produced by CloudWatch systems should be explicit in their naming about the type of data they process. For instance, CloudWatch Metric Streams can produce three different output formats (though two of those are different OTLP versions). This is different from EMF, which is another CloudWatch metric encoding scheme that may be found in, for example, log records or S3 objects.

[constanca-m]

For instance, CloudWatch Metric Streams can produce three different output formats

Thanks @Aneurysm9, I have updated the issue to only be applicable to cloudwatch metric streams for now. In the future, we should add encodings for cloudwatch logs and the other types of CW metrics.

[atoulme]

I can sponsor this component.

[axw]

Initial skeleton PR: #38388

[axw]

#38407 is merged. Once #38408 is done I think we can look at progressing this to Alpha stability.

[axw]

#38704 is up to bump to Alpha.