Service goals
With ThruPut Manager AE, each job inherits its goals from the category to which it belongs, and then TM AE manages the job to meet its goals. This is in contrast to managing the group of jobs, belonging to a category, to meet the goals of the category. The TM AE algorithm ensures that no job languished excessively.
This is one of the major differences between the ThruPut Manager approach and the WLM approach.
With ThruPut Manager AE, for a particular category of batch work, you can specify goals, or service times, in simple wall clock time as follows:
- A value, in minutes and seconds, by which you’d target a job to be selected for initiation (the Target service time);
- A value, in minutes and seconds, within which you’d consider it acceptable for the job to be selected for initiation (the Acceptable service time);
- A third value, in minutes and seconds, after which you’d consider it critical that the job be escalated if it hasn’t yet been selected for initiation (the Critical service time). This value is typically used only for the most important workloads, so these jobs are given priority for selection even when the system is busy.
These service times are datacenter commitments to provide service to a job (to initiate it) within a specific time after it enters the system. SLM also uses these values to decide how fast to age the jobs in the queue – the smaller the value the
faster they age (More on this in Queue Management).
Related topic