You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
From the report generated from http://sparklens.qubole.com/ , I see the ideal executor plot where it plot "the minimal number of executors (ideal) which could have finished the same work in same amount of wall clock time"
I am curious what are the formulas, equations for such plot. If you can give me some explanation on how you guys approach it, that would be great. Thanks!
The text was updated successfully, but these errors were encountered:
Thanks for the wonderful words. The calculation you are referring to comes from simulation. It works just like the completion time vs efficiency graph at different executor counts. Instead of simulating the complete application, here we simulate each spark job individually. To find the minimal number of executors required for a particular job, we do a binary search between 1 and total number of executors. This graph is basically trying to show if autoscaling would be useful and second to judge how a given autoscaling policy is doing compared to "ideal" autoscaling.
Hi,
First of all, amazing project!
From the report generated from http://sparklens.qubole.com/ , I see the ideal executor plot where it plot "the minimal number of executors (ideal) which could have finished the same work in same amount of wall clock time"
I am curious what are the formulas, equations for such plot. If you can give me some explanation on how you guys approach it, that would be great. Thanks!
The text was updated successfully, but these errors were encountered: