-
-
Notifications
You must be signed in to change notification settings - Fork 193
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
The time a Cloudlet waits after being preempted is not computed #151
Comments
Hey @manoelcampos , |
Hey @krupaJari |
Do you have anything in mind already on how to design a solution? It would be great if you provide your thoughts abouth the discussion on the issue above. |
The regular Cloudlet workflow is as follow:
CloudletSchedulerTimeShared
, it will share the CPU with other concurrent Cloudlets. This way, multiple Cloudlets are executed preemptively (i.e., interchangeably).However, the time a preempted Cloudlet is supposed to wait until be scheduled again is not computed.
The oversimplified
CloudletSchedulerTimeShared
doesn't actually perform preemption, but just reduces the amount of MIPS a Cloudlet can use when there are more Cloudlets than CPU cores.This way, it enables all Cloudlets to run at the same time (which is not possible with current processor technologies). Check issues #27, #33 and #58 for more details.
When a CloudletSchedulerSpaceShared is used, when there are fewer CPU cores than Cloudlets, the next Cloudlet just starts executing when a previous one completely finishes its execution. This wait time is not computed either.
The
Cloudlet.getWaitingTime()
method returns the time the Cloudlet waited before starting executing, not the intermediate wait time after the Cloudlet started executing but was descheduled (preempted) for some reason. Before the implementation of Google Cluster Data readers (#149), a Cloudlet was never descheduled, but now this is a reality (as it is in real CPU scheduling).It has to be assessed if the
Cloudlet.getWaitingTime()
will be used just to return the time the Cloudlet waited to be scheduled for the first time (as it is now) and a new method such asgetTotalWaitingTime()
will be introduced that adds up the intermediate waiting times. Possibly the first method should be renamed togetStartWaitingTime()
.The text was updated successfully, but these errors were encountered: