Scheduling criteria

User-oriented scheduling criteria

Turnaround time

Interval of time between the submission of a process/job and its completion (includes actual execution time plus time spent waiting for resources, including the processor).

• the appropriate measure for a batch job.

• should be minimized.

Waiting time

Sum of periods spent by a process waiting in the ready state.

• should be minimized.

Response time

Time from the submission of a request until the response begins to be received.

• appropriate measure for an interactive process.

• should be minimized.

• but also the number of interactive processes with acceptable response time should also be maximized.

Deadlines

Time of completion of a process.

• percentage of deadlines met should be maximized, even subordinating other goals.

Predictability

How response is affected by the load on the system.

• A given job should run in about the same amount of time and at about the same cost regardless of the load on the system.

System-oriented scheduling criteria

Fairness

Equality of treatment.

• In the absence of guidance, processes should be treated the same, and no process should suffer starvation.

Throughput

Number of processes completed per unit of time.

• measures the amount of work being performed by the system.

• should be maximized.

• depends on the average lengths of processes but also on the scheduling policy.

Processor utilization

Percentage of time that the processor is busy.

• should be maximized (especially in expensive shared systems).

Enforcing priorities

Higher-priority processes should be favoured.

As referred to before, it is impossible to satisfy all criteria simultaneously.