A series of questions related to Predictive Events |
Q1 - What is a Predictive Event Ans - The Prediction feature gives early warnings of certain exceptional situations and is used to issue warnings if there is an aggressive trend towards the threshold. While it can be used in a wide number of scenarios, it is most effective in capacity-type scenarios, especially for those metrics which exhibit clear hourly patterns. Q2 - Please elaborate the actual scenarios where it calculates the predictive event on generic situations, as we need to understand how the predictive event is generated on various occasions. Ans - Predictive Events can be generated for any attributes. If the behavior of the component being monitored is constant for weeks and months, you should go for Signature Thresholds which will let the you know when there is a different behavior than the normal behavior. More than prediction, this gives an early indication based on which customer has to act. Q3 - Need Information on the delta, for the discussed scenario where event is generating ~57% which is less than confirmed value 60% as compared to the configured Absolute value 80% Ans - The system learn the slope will meet the absolute threshold within 3 hours. That is the reason it generates the prediction and closes once when the promising slope is no more there. Q4 - Please share the significance for the desired alarms which are out of the box. Eg: a) Message1 : “ Predicted to alarm in the next hour: “ Filesystem % space used”. Eg: b) Message2 : “ Predicted to alarm in the next 2 hour: “ Filesystem % space used”. Ans - The time period with in which the prediction will get generated is 3 hours max. Depending on at what time the prediction will convert into actual event, the time range is quoted. Like with in 1 hour or 2 hour or 3 hour. Q5 - Please share the sliding window concept associated with predictive events Ans - For the predictive event to get generated and for the same to remain, there should a linear regression and at any point of time the data point shows a saturation (meaning value are not linearly regressing but it tends towards a straight line), then the predictive event is no more valid. Sliding window here is the last n number of data points which is there in the buffer against which the calculation is made. Buffer will constitute of 20 minutes of data values (by default). Q6 - What is a Signature Threshold Ans - Signature thresholds are more appropriate for metrics that are more transient, such as response time, packet per second, and so on, where the data can exhibit big swings against a generally upward/downward trend. |