Calculating and monitoring critical paths

For each operation defined as critical, a specific critical path is calculated in the dependency network during daily planning processing. Starting from the target operation, for each level of predecessors, the process chooses the most critical one and includes it in the critical path. Among all of the internal and external predecessors, the most critical one is planned to be the predecessor with the latest end time. Calculating the critical path starts from the target operation and moves backward among its predecessors; it ends when an operation without predecessor is reached.

The deadline of a critical operation should not be later than the deadline of the application to which the operation belongs. If this is not the case, message EQQA124I SPECIFIED OP DEADLINE LATER THAN APPLICATION DEADLINE is issued, along with the short message TIME INCONSISTENT.

The planned end time is calculated as the output of the HCL Workload Automation for Z planning processing, which is based on the information you set for the operation or occurrence (for example the IA time, duration, and deadline). The workstation parallel servers, open intervals, special resources and their availability are also taken into account. For this reason, the more accurate the application definition, the more accurate the critical path calculation.

To reduce the performance effort to a minimum, the estimated start and end times are updated only in a limited number of cases:
  • When a late condition occurs and the accumulated delay determines the need for a critical path recalculation, the estimated times of the entire branch in the critical network identified by the late operation are updated.
  • When an operation located on a critical path completes, the completion time is compared with the estimated end time and a delay or an advance is propagated among the successors on the critical path.
  • When an occurrence belonging to a critical network is dynamically added to the current plan, all its internal operations will have their estimated times updated to consider the internal dependencies.
In all other cases, to update the estimated times in the critical network you must run a daily planning job EXTEND or REPLAN.

For each critical job, you are provided with a percentage that indicates the confidence that the critical job will meet its deadline. The confidence factor is calculated as the normal cumulative density function: it is the probability that the job will end within its deadline, calculated by using a Gaussian function, where the estimated end time is the mean and the estimated end variance is the standard deviation.

To reduce the performance effort to a minimum, the confidence factor is calculated only when the estimated times are updated. Consequently, the value of the confidence factor always corresponds to the latest calculation.

When a critical job becomes late, its confidence factor is automatically set to 0. When you dynamically set a job as critical, the value for its confidence factor is set to 50.

Note: Keep in mind that for shadow jobs involved in a critical path, the estimated duration is always calculated as one minute.

To have HCL Workload Automation for Z update your estimates in the application description database with actual run values, see Using duration feedback options.

While the plan is running, the scheduler checks if any predecessor of a critical job is starting to delay and monitors the critical paths that are consuming their slack time. In particular:
  • The scheduler maintains a hot list for critical jobs, monitoring any critical job predecessor that starts to delay in one of the following conditions:
    • It is late, meaning that it did not start by its latest start time
    • It is long running, meaning that it is running longer than its estimated duration
    • It ends in error
    • It is suppressed by condition
    To do this, the scheduler applies the same internal logic that is used to monitor alert conditions.

    Before recalculating a critical path, the scheduler estimates new start and end times for all the successors of the job that started to delay.

  • When any job in a critical path completes or a dynamic update changes the path, before recalculating the critical path the scheduler estimates new start and end times for the successors of the changed job.
  • When the estimated start and end times determine a critical path change, the scheduler updates the current plan.

The deadlines of the operations in a critical job network affect the accuracy of the dynamic critical path handling, To optimize this, set these deadlines to the same or later than the critical job deadlines.

You can monitor critical paths by selecting option 7 (Critical Jobs) from the CURRENT PLAN AND STATUS INQUIRY panel. You get a list of all the critical target operations included in the current plan. Select the operation you want, to display the list of operations belonging to the corresponding critical path in BROWSING THE CRITICAL PATH . For details about option 7 (Critical Jobs), see CRITICAL JOBS.

You can monitor critical paths also through the TEP monitor. For details, see Enabling monitoring on HCL Workload Automation for Z.