A DevOps team manages an API running on-premises that serves as a backend for an Amazon API Gateway endpoint. Customers have been complaining about high response latencies, which the development…

Question

Options

• AInstall the CloudWatch agent server side and configure the agent to upload relevant logs to
• BEnable AWS X-Ray tracing in API Gateway, modify the application to capture request segments,
• CEnable AWS X-Ray tracing in API Gateway, modify the application to capture request segments,
• DModify the on-premises application to send log information back to API Gateway with each
• EModify the on-premises application to calculate and upload statistical data relevant to the API

• A
  76% (19)
• B
  16% (4)
• D
  4% (1)
• E
  4% (1)

Explanation

Explanation

Installing the CloudWatch agent on the on-premises server (A) allows logs and metrics to be collected and uploaded to CloudWatch passively, without adding any overhead to the API request/response cycle itself. Enabling AWS X-Ray tracing in API Gateway and modifying the application to capture request segments (C) provides end-to-end distributed tracing, allowing the team to pinpoint exactly where latency is occurring - whether in API Gateway, the network, or the backend application - without adding measurable latency to requests.

Why the distractors are wrong:

• B appears nearly identical to C but likely includes an incorrect configuration detail (such as sending traces to the wrong destination or adding synchronous overhead)
• D is wrong because sending log data back through API Gateway with each request adds extra payload and processing, which would increase latency rather than diagnose it
• E is wrong because calculating and uploading statistical data from the application introduces additional compute and network overhead per request, potentially worsening the latency problem

Memory Tip: Think "passive observation only" - the correct answers collect data alongside the request flow (agent-based logging + X-Ray tracing), never within the request path. If an option routes diagnostic data back through the API or adds per-request computation, it violates this principle and should be eliminated.

No community discussion yet for this question.