Starting from 930, **Workload Service Assurance**, aka **Dynamic Critical Path**, has been significantly enhanced in determining the **High Risk Level for a Critical Job**.

The original issue was to improve precision in determining the Risk Level. During these years some specific enhancements have been implemented in this direction.

The most critical area was the calculation of **Estimated Start and End Times**. The two main "actors" for determining Estimated times are** deadlines **and **durations**.

Before 930 a lot of work has been done on deadlines, starting from the **BATCHOPT** parameter **IGNOREDEADL**, and ending with the optional deadline introduced by TWSz 910. These enhancements allow late processing of the critical networks to be completely based on the deadline of the critical job. This way we avoid the risk of false alarms that might be generated by other, not critical, operations going late respect to their own deadlines.

Starting from 930 we have **variable deadlines** and **variable durations**: for an operation we can specify different deadlines and durations for different run cycles. This way, time calculations are more precise and the monitoring results are more realistic in the day by day activity.

One of the most important enhancements offered by 930 is the introduction of the** Variance in Duration**. Each time an estimated duration is updated by the effect of the **Limit Feedback** and the** Smoothing Factor**, variance is calculated and stored in the record. This variance is used by each CP create/extend batch to calculate the variance in the Planned Start and End Times. and by The Critical Path handler task to calculate (and keep up to date) the variance in the Estimated Start and End Times. This variance is used by CPH task to calculate the **Confidence Factor for a Critical Job**, that is the probability that the critical job meets its deadline.

Confidence Factor is calculated by means of the **Gaussian Distribution** of the critical job's possible end times, respect to the interval determined by the Estimated End Time variance. It provides the probability that the Actual End Time occurs before the deadline (the yellow area in the picture below).

Looking at the Gaussian curve, all the values for deadline that are out of the "bell" will determine the value 0% or 100% for Confidence Factor depending on the fact that deadline is before or after the Estimated End Time. All the other values will be calculated by the Gaussian Function.

There is a way to link the High Risk Level to the Confidence Factor for a Critical Job: by means of **JTOPT **parameter **RISKCONFIDENCE**. By default (i.e. when the parameter is not specified) the Risk Level for a Critical Job will be set to "High" according to the old criterium: when the Estimated End Time goes beyond the deadline. When a value x is specified for the keyword RISKCONFIDENCE, the risk level will be High when the Confidence Factor is smaller than x.

It is easy to understand that the old behavior is determined by the value 50 for the keyword RISKCONFIDENCE.

By means of this parameter we can decide more precisely the criterium for the High Risk Level to be triggered. In fact working only with the ON/OFF to determine the high risk, as the feature has done so far, was becoming a more and more restrictive use for a so powerful feature!