Adaptive Threshold - FAQs

1. Is the Adaptive Thresholds feature available for trial users?

Yes. The adaptive thresholds feature is enabled by default and is available for trial users.

2. After enabling the adaptive thresholds feature, will there be any delay in raising alerts?

To raise alerts using the adaptive thresholds feature, we need a minimum of 3 days of monitor data for the AI to compute thresholds. If three days of monitor data is readily available, you can instantly receive alerts. If not, it will take three days for the data collection post which you will receive alerts.

3. Is it possible to enable Adaptive thresholds for custom CPU utilization monitor?

Our Adaptive Threshold feature is in its beta stage and is limited to the following three monitors:

  • CPU Utilization
  • Memory Utilization
  • Response Time

However, we will bring in the adaptive threshold feature for custom monitors in our upcoming versions.

4. What happens if the user disables the Adaptive Threshold feature?

We will retrieve all our adaptive thresholds associated with the device, monitor and template levels and, restore previously configured manual thresholds if any.

5. If the adaptive threshold feature is disabled and then enabled later, will the learning of ML be removed?

Yes. Once the adaptive threshold feature is disabled, all its learnings (Model data) are removed. A new model will be created once the adaptive threshold feature is enabled again.

6. Which ML algorithm is being used in the adaptive threshold feature?

We use Manage Engine's in-house developed ML algorithm that uses multiple techniques such as regression, STL Decomposition cascaded with Spectral and Regression models, ETS (Explicit handling of Trend and Seasonality) and ARIMA (Auto-Regressive Integrated Moving Average).

7. What happens if I apply for Standard or Lite edition license after enabling the adaptive Threshold feature?

Once you apply for the standard/lite edition, the Adaptive threshold feature will automatically be reverted, and you can no longer enable it.

8. Do read-only users have the privilege to view or edit AT settings?

No. Read-only users can view Adaptive threshold settings but, they cannot update or associate them.

9. What is the deviation in Adaptive Threshold values?

Using Machine Learning, we will forecast the value of the metric for Twenty-four hours. These forecasted values will act as a baseline. The deviation values which are configured in the UI will be added to these baseline values to know the exact threshold value. If a user wants narrowed thresholds, they can decrease the deviation values for each severity. These user-assigned deviation values will indirectly give the user some control in defining stricter or lenient thresholds based on their requirements.

10. Can a user see the forecasted threshold values for the forthcoming hours?

No. For now, this information is stored within the database itself. However, we are planning to add this feature to our upcoming versions.

11. Is it possible to configure different deviation values for different groups of servers?

No. Right now, for one installation, only one set of deviation values will be there for a monitor.

12. How can I use the Adaptive threshold feature only for a particular set of devices?

This can be achieved by opening the device template, enabling the adaptive threshold feature and associating the same with the devices. You can also achieve the same in the performance monitoring page under the monitoring option of the settings tab. (Settings >> Monitoring >> Performance Monitors)

 
 Pricing  Get Quote