What should the SysOps administrator do to meet this requirement?
Create an Amazon CloudWatch Contributor Insights rule on the log data.
Create a metric filter on the log data.
Create a subscription filter on the log data.
Create an Amazon CloudWatch Application Insights rule for the workload.
Explanations:
Amazon CloudWatch Contributor Insights is designed to analyze log data and identify key contributors to issues or anomalies, but it does not directly provide the ability to calculate specific statistical metrics like p90 latency over time.
A metric filter allows you to extract numeric values from log data and convert them into CloudWatch metrics. By applying a metric filter to the application latency field, you can then create a metric and use CloudWatch to calculate the p90 percentile over time.
A subscription filter forwards log data to a destination such as Amazon Kinesis or Lambda, but it doesn’t provide the ability to calculate or monitor statistics like the p90 latency.
Amazon CloudWatch Application Insights is used to monitor applications, detect anomalies, and provide insights into operational issues. However, it doesn’t focus on calculating percentile statistics from log data directly.
I systematize that the answer is:
Create a metric filter on the log data.