SuiteCloud Processors Priority Elevation and Processor Reservation (Advanced Settings)
Priority Elevation
Priority elevation enables you to automatically increase the priority of each low or standard priority job after a specific time interval. The time interval starts when the job is submitted. You do not need to use these settings in most cases. Because of this, priority elevation is disabled by default. However, you may need to use them if lower priority jobs experience excessive wait times.
Priority elevation only impacts lower priority jobs with a wait time greater than the time interval indicated. If a lower priority job is sent to the processor pool before the time interval is over, that job is processed with its original priority.
To access the priority elevation settings, go to Setup > Preferences > SuiteCloud Processors.
There are three predetermined settings listed:
-
No Priority Elevation: The default setting
-
Moderate Priority Elevation:
-
If a low priority job is still waiting after four hours, it is elevated to standard priority
-
If a standard priority job is still waiting after four hours, it is elevated to high priority
With this option, if priority elevation is applicable, the system elevates low priority jobs to high priority jobs after eight hours. Specifically, the jobs in this scenario are elevated to standard priority after four hours and then elevated to high priority after another four hours.
-
-
Intensive Priority Elevation:
-
If a low priority job is still waiting after one hour, it is elevated to standard priority
-
If a standard priority job is still waiting after one hour, it is elevated to high priority
With this option, if priority elevation is applicable, the system elevates low priority jobs to high priority jobs after two hours. Specifically, the jobs in this scenario are elevated to standard priority after one hour and then elevated to high priority after another hour.
-
-
Custom Priority Elevation: This option enables you to specify a custom time interval for priority elevation from the Time Interval dropdown list.
The Time Interval field specifies the time interval set for priority elevation. When Custom Priority Elevation is selected, this field is enabled for editing. Otherwise, the field displays a value that corresponds with the option selected, but it cannot be edited.
Click Advanced at the top of the page to access Custom Priority Elevation and Time Interval.
Processor Reservation
Processor reservation enables you to reserve a specified number of processors for high priority jobs. You do not need to use these settings in most cases. Because of this, processor reservation is disabled by default.
Processor reservation is available for SuiteCloud Plus accounts only.
To access the priority elevation settings, go to Setup > Preferences > SuiteCloud Processors. Click Advanced at the top of the page.
When you select Enable Reservation, you can reserve all but one of your available processors from the Number of Processors Reserved dropdown list. If a high priority job is submitted, it is sent to the processor pool if there is at least one processor available. If a standard or low priority job is submitted, it is sent to the processor pool only if there are more processors available than the number reserved. For example, you have 10 processors reserved out of 25 total processors. A standard or low priority job is sent to the processor pool if there are at least 11 processors available. If there are 10 processors or fewer available, the lower priority job must wait.
Changes to the Number of Processors Reserved apply to all jobs that have not yet started. This can have an immediate impact on map/reduce scripts since each stage is processed by a minimum of one job. If there is a high priority map/reduce script instance (task) executing and this setting is changed, the new value is applied to all jobs for this task that have not yet started. This includes jobs that may be created from yielding.
Processor reservation decreases the number of processors available for standard and low priority jobs. Therefore, it can reduce the throughput of these jobs. The Reuse Idle Processors setting temporarily releases reserved processors that have not been used in the past 24 hours. This increases the number of processors for lower priority jobs.
When Reuse Idle Processors is enabled, it initiates an hourly recurring audit. The system uses the data collected to determine whether to release reserved processors. After reserved processors are released, the audit data is also used to determine whether the system needs to increase reserved processors.
If the system decreases or increases the number of reserved processors, additional decreases are not made for 24 hours. However, additional increases (up to the selected limit) can still be made after each hourly audit. This process continues if Reuse Idle Processors is enabled.
The system analyzes the following data points during this process:
-
a = total number of processors available
-
b = the value you set for Number of Processors Reserved (the maximum number of reserved processors; this number does not change)
-
c = maximum number of jobs concurrently processed in the last 24 hours
-
d = maximum number of high priority jobs that concurrently waited more than a minute in the last hour
-
e = current number of reserved processors (this number can change if Reuse Idle Processors is enabled)
|
If |
Then |
|---|---|
|
c is less than a This means that some reserved processors were not used. |
The number of processors available to lower priority jobs is increased by a – c (up to the value of a). |
|
e is less than b AND d is more than 0 |
The number of reserved processors is increased by d (up to the value of b) The system cannot increase the number of reserved processors over the limit set for Number of Processors Reserved. In other words, e cannot be greater than b. |