Understanding P90 Duration
What is P90 Duration?
P90 duration is a statistical metric representing the 90th percentile of response times for a given data set.
This means 90% of the observed response times are equal to or faster than this value, while the slowest 10% are excluded.
Benefits of P90 Duration
🧑💻 Focuses on User Experience: P90 highlights the response times experienced by the vast majority of users, excluding rare extreme cases.
📉 Reduces Noise from Outliers: Unlike averages, P90 is less affected by a small number of unusually slow requests, making it a more stable metric for performance monitoring.
🎯 Better for Service Level Objectives (SLOs): Many organizations use P90 or P95 metrics in their SLOs to ensure consistent performance for the majority of users.
Example
Let's imagine the following response times (sorted by duration):
Request | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
---|---|---|---|---|---|---|---|---|---|---|
Time (ms) | 100 | 110 | 120 | 130 | 140 | 150 | 160 | 170 | 180 | 2000 |
- Average Duration:
(100 + 110 + 120 + 130 + 140 + 150 + 160 + 170 + 180 + 2000) / 10 = 426ms
- P90 Duration: Sort the data
100, 110, 120, 130, 140, 150, 160, 170, 180, 2000
, then take the value at the 90th percentile:180ms
As shown, the average is heavily influenced by the single outlier (2000 ms), while P90 provides a clearer picture of typical performance by focusing on the majority of requests.
Conclusion
Whether you are tracking endpoint trends, comparing deployments, setting SLOs, or conducting performance tests, P90 helps you focus on the majority of user experiences while filtering out extreme outliers.
By leveraging P90, you can ensure your application delivers consistent, high-quality performance to your users.
If you have any questions about P90 duration or how it’s used in Hud, contact support at [email protected].
Updated 17 days ago