Feature request: Client-Side sampling or similar mechanism to avoid throttling for traces

[slavaschmidt]

Use case

Context

We are currently evaluating AWS Powertools as a potential replacement for our custom X-Ray trace publisher. Our existing solution is tailored for high-throughput batch processing workloads, where we frequently encounter throttling issues with AWS X-Ray during peak load periods.

Use Case

In our current setup, we mitigate X-Ray throttling by implementing a custom sampling mechanism using SQS. This allows us to decouple trace generation from trace submission and apply post-processing sampling logic before sending data to X-Ray. However, AWS Powertools does not currently offer a way to configure or intercept trace processing in a way that would allow us to implement similar throttling mitigation strategies.

Solution/User Experience

We would greatly benefit from a mechanism in Powertools that allows more control over how traces are handled before being sent to X-Ray. Two possible approaches could include:

• Configurable Sampling Logic: Allow users to define custom sampling strategies (e.g., probabilistic, rate-based, or dynamic sampling).
• Pluggable Trace Destination: Provide an option to redirect trace data to a custom destination (e.g., SQS, Kinesis, or a local buffer) for deferred or filtered submission.

We are open to any other solution that would enable similar flexibility in managing trace throughput under high-load conditions.

Thank you for considering this enhancement!

Alternative solutions

Acknowledgment

• [x] This feature request meets Powertools for AWS Lambda (Java) Tenets
• [x] Should this be considered in other Powertools for AWS Lambda languages? i.e. Python, TypeScript, and .NET

Future readers

Please react with 👍 and your use case to help us understand customer demand.