Skip to content
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

estat standard deviation looks wrong #46

Open
sebroy opened this issue May 14, 2020 · 0 comments
Open

estat standard deviation looks wrong #46

sebroy opened this issue May 14, 2020 · 0 comments

Comments

@sebroy
Copy link
Contributor

sebroy commented May 14, 2020

Consider the following output from the nfs script:

   microseconds                                          v3/read, cached
value range                 count ------------- Distribution -------------
[20, 30)                        7 |@
[30, 40)                       12 |@@
[40, 50)                       49 |@@@@@@@
[50, 60)                       21 |@@@
[60, 70)                       32 |@@@@@
[70, 80)                       44 |@@@@@@@
[80, 90)                       31 |@@@@@
[90, 100)                      15 |@@
[100, 200)                     60 |@@@@@@@@@
[200, 300)                      6 |@
[300, 400)                      4 |@
[400, 500)                      1 |@

   microseconds                                           v3/write, sync
value range                 count ------------- Distribution -------------
[300, 400)                    398 |@
[400, 500)                   3545 |@@@@
[500, 600)                   5085 |@@@@@@
[600, 700)                   4298 |@@@@@
[700, 800)                   3261 |@@@@
[800, 900)                   2213 |@@@
[900, 1000)                  1732 |@@
[1000, 2000)                 9806 |@@@@@@@@@@@
[2000, 3000)                 3788 |@@@@
[3000, 4000)                 1929 |@@@
[4000, 5000)                 1079 |@@
[5000, 6000)                  599 |@
[6000, 7000)                  247 |@
[7000, 8000)                  115 |@
[8000, 9000)                   65 |@
[9000, 10000)                  35 |@
[10000, 20000)                 41 |@

   microseconds                                        v3/read, uncached
value range                 count ------------- Distribution -------------
[400, 500)                      1 |@
[500, 600)                     10 |@@@@@@
[600, 700)                     33 |@@@@@@@@@@@@@@@@@@@@@
[700, 800)                     11 |@@@@@@@
[800, 900)                      2 |@
[900, 1000)                     1 |@
[1000, 2000)                    2 |@
[2000, 3000)                    1 |@
[4000, 5000)                    1 |@

                                       iops(/s)  avg latency(us)       stddev(us)  throughput(k/s)
v3/read, cached                              56               88             3229              295
v3/write, sync                             7647             1439          1726695            59783
v3/read, uncached                            12              766           244436               74


                                       iops(/s)  throughput(k/s)
total                                      7724            60073

Looking at the v3/write, sync operations, the latency standard deviation (1,726,695) is higher than the ceiling of the highest latency bucket (20,000), which is mathematically impossible.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Development

No branches or pull requests

1 participant