-
Notifications
You must be signed in to change notification settings - Fork 478
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
support percentile time in uftrace report #1982
Comments
Thanks for the report. I think this feature would be very useful. The only concern I have is that calculating those percentiles for entire functions in the record might take a long time as it requires sorting each record. I would like to hear how @namhyung thinks about it. |
We might have to handle the performance issue by creating multiple threads. |
Well, I think it can just append the time data and sort it once at the end. The sorting can be multi-threaded. |
Hello, I would like to take this issue or any other you would recommend, for learning purpose. Would you have any useful information you could give me ? |
Hi, it'd be nice if you could take it. You could start with a simple approach first, not considering multi-threaded cases. But it should be optional and only enabled if one of percentile fields are used. |
I am writing to propose an enhancement to the uftrace reporting capabilities that I believe would be highly beneficial for users who analyze function tail latencies.
Percentile metrics, such as the 90th, 95th, 99th, or 99.9th percentiles, are widely used as key indicators when measuring tail latency. While it is possible to extract these percentiles through
uftrace script
, having them natively available in theuftrace report
would enhance usability and provide immediate insight into latency distributions.Specifically, I suggest adding the following fields to the report:
total-p90
,self-p90
,total-p95
,self-p95
,total-p99
, andself-p99
.The text was updated successfully, but these errors were encountered: