Which percentiles are used for detecting slowdown events in Dynatrace?

In Dynatrace, the use of the 50th (median) and 90th percentiles for detecting slowdown events is particularly effective because these percentiles provide a balanced view of performance data. The 50th percentile represents the median response time, indicating the point at which half of the requests are faster and half are slower. This is crucial for understanding typical user experiences.

The 90th percentile represents a more extreme case, identifying performance issues that affect a significant portion of users—specifically, the top 10% of requests that are experiencing the slowest response times. By focusing on these two percentiles, Dynatrace can effectively pinpoint slowdowns that impact user experience without being skewed by outliers or extreme values, which might misrepresent overall performance if only average values were used.

This combination helps organizations quickly identify and address performance issues, ensuring a smooth experience for users by addressing response times that are worse than average while still maintaining a focus on the typical user experience.