Self Tuning Threshold (STT) monitor is a monitor type in the Operations Manager 2007 library of monitor types. Instead of a fixed threshold, STT monitors have self-tuning thresholds which are learned from the environment and updated over time. This post is about the concepts related to STT monitors.


Baseline is a set of average and standard deviation values. The low and high threshold values are calculated based on the values in the baseline. Baseline is computed from the performance data in the environment.

Business Cycle:

A Business Cycle is a recurring time period during which a specific performance counter’s values follow a similar pattern. For example, think about the number of simultaneously logged on users in an organization. This number will increase around 9AM and decrease around 6PM during week days. During the weekends, the number will be almost constant since only few people log on. Here we can tell that the business cycle is a week since the numbers follow the same pattern each week.

Initial Learning Period (in Business Cycles):

Initial Learning Period is the time period during which the initial baseline is established. Initial Learning Period is specified in business cycles. Alerts are not generated during this time. After the initial learning period, the incoming performance data is compared against the computed thresholds.


Sensitivity is the level of acceptable tolerance for deviation in amplitude from the baseline. Sensitivity value is used to calculate the thresholds. Lower sensitivity allows higher deviations from the baseline to be within the thresholds. The graphs below show how the low and high sensitivity values affect the calculated thresholds.






Learning Rate:

Learning rate is the relative importance the learning algorithm gives to recent observations compared to the historical data. Fast learning means that the baseline will look more like recent observations.






 Time Sensitivity:

Time Sensitivity is the level of acceptable tolerance for deviation in timeliness from the norm. Higher time sensitivity allows bigger deviations in timeliness from the norm to be acceptable. For example, let say the number of logons per minute in an organization reaches peak at 9:00 AM since most employees come to work at 9:00 AM. Let’s say, one day most of the employees are late by 30 minutes because of heavy snow. In this case, the new peak has shifted half an hour and now is at 9:30 AM. Time sensitivity allows us to make these kinds of shifts to be acceptable or not.  Higher time sensitivity value allows bigger shifts to stay within the thresholds