問題文
A dashboard shows the average response time of an internal service at 120 ms, well within the 500 ms target. Users report that the service is regularly unusable. Which measurement should be added to reconcile the two?
選択肢
- The availability of the service, since it is a separate indicator and a service that is briefly down still reports a low average.
- The number of requests per second, because a higher request rate raises the average response time and comparing the two series shows whether the reported unusability coincides with the periods of highest load on the service.
- The error rate, since failures do not appear in the response time and a failed request is what users experience as an unusable service.
- The distribution of response times, because an average hides a tail in which a share of requests take far longer.