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
The idea here is to try to make one visualization that packs a lot of information as to what's happening in a provisioning simulation. This is not to the exclusion of graphing metrics or calculating values that matter, but intended to let us "understand" how a particular provisioning approach works in a situation, or how two approaches differ.
Weird sketch here:
The idea is to show the throughput of tasks on the left side, and the worker capacity on the right side.
highlights the things we want to avoid -- idleness and pending -- but in a way that we can see what portion of the overall load those represent
shows the rate of task execution by the angle of the lines of shading on the left
shows the rate of worker churn by the angle of the lines of shading on the right
shows total concurrent tasks by the width of the left side
shows total amount of worker capacity by the width of the right side
shows it all in a way that we can compare across time and see things like the worker surge that follows a big spike in tasks
I think the left side would be the concatenation of all tasks, from youngest to oldest, with each task's width given by the remaining duration, and the color by its status (yellow=pending, orange=running) and identity (subtle shade differences based on taskId)
the right side would be the concatenation of all workers, with the width given by the worker's total work per second, colored fractionally purple for running and blue for idle (or starting up), with subtle shade differences for worker name
I suppose different colors for starting up and idle would be good
The text was updated successfully, but these errors were encountered:
@djmitche commented on Mon Aug 03 2020
The idea here is to try to make one visualization that packs a lot of information as to what's happening in a provisioning simulation. This is not to the exclusion of graphing metrics or calculating values that matter, but intended to let us "understand" how a particular provisioning approach works in a situation, or how two approaches differ.
Weird sketch here:
The idea is to show the throughput of tasks on the left side, and the worker capacity on the right side.
I think the left side would be the concatenation of all tasks, from youngest to oldest, with each task's width given by the remaining duration, and the color by its status (yellow=pending, orange=running) and identity (subtle shade differences based on taskId)
the right side would be the concatenation of all workers, with the width given by the worker's total work per second, colored fractionally purple for running and blue for idle (or starting up), with subtle shade differences for worker name
I suppose different colors for starting up and idle would be good
The text was updated successfully, but these errors were encountered: