An on-premises backend API serves requests for an API Gateway endpoint. Customers report high response latency, which API Gateway latency metrics confirm. To diagnose the issue, the team needs to collect relevant telemetry without adding extra latency to requests. Which two actions should they take? (Choose two.)

Installing the CloudWatch agent on the server side and configuring it to send logs to Amazon CloudWatch provides detailed insights into the on-premises server's performance, resource utilization, and application-specific logs, which are crucial for diagnosing latency issues. This process is asynchronous and does not add significant latency to the request path. Enabling AWS X-Ray tracing in API Gateway and modifying the application to capture request segments, then using the X-Ray daemon, allows for end-to-end request tracing without adding latency to the request itself, as the daemon buffers and uploads segments asynchronously. Modifying the application to send log information back to API Gateway or publishing aggregated metrics directly would add latency to each request, which contradicts the requirement. Enabling X-Ray tracing and modifying the application to upload segments during each request would also add latency.