How to ensure the standard deviation threshold profile is working as configured in DX NetOps CA Performance Management (CAPM)
book
Article ID: 272080
calendar_today
Updated On:
Products
CA Performance Management - Usage and AdministrationDX NetOps
Issue/Introduction
We are unable to determine whether the alarms generated are genuine because we are not sure what the baseline and the 4.0 standard deviation values are.
Threshold Profile Configuration and Alarm Verification
We have configured a threshold profile with the following parameters:
Metrics:
Bits in per second
Bits out per second
Thresholds:
Above 4.0 standard deviation for Bits in per second for a 1200-second duration in a 1200-second window.
Above 4.0 standard deviation for Bits out per second for a 1200-second duration in a 1200-second window.
Environment
DX NetOps Performance Management All Releases
Resolution
Guidelines for Monitoring Bits In/Out per Second with Baseline and Standard Deviation
Viewing Baselines:
You should be able to add the baseline to the views showing Bits In per second and Bits Out per second. This will allow you to see what the baseline is at any given time.
Reference Documentation:
For detailed information on how both baseline and Standard Deviation are calculated, please refer to the following sections in our TechDocs:
You can extend the Metric Family (MF) and enable Standard Deviation (stddev) and baselines for any metric that doesn't already have them. However, Bits per second in/out already has baseline and stddev enabled.
Baseline and Projection Calculations:
We calculate hourly and daily baselines (mean/min/max/stddev) and projections if enabled.
Thresholding stddev checks are based on the baseline stddev for the hour or day, depending on the available data.
Note that the stddev in the rate (as-polled table) reflects only one data point and is therefore not useful for thresholding.
Trend Graphs:
If you are looking to create a trend graph of the polled value and baseline stddev for the same metrics, keep in mind:
Stddev may not be sufficient due to variance from item to item; some items may have large stddev, while others may have very small stddev.
Applying a 4x stddev may be fine for some items but too small for others.
To address this, consider using a combination of % of baseline and stddev. This ensures that items with very small stddev will require significant movement in value before triggering the stddev check.
Using Custom Views:
Use IM Custom Trend as the Out-of-the-Box (OOTB) views will not provide the deviation.
Click on the baseline metrics checkbox to make the baseline mean/min/max/stddev visible.
Plot the read as-polled metric via the baseline mean and Standard Deviation to visualize the deviation.
By following these guidelines, you can effectively monitor and analyze Bits In per second and Bits Out per second using baseline and Standard Deviation metrics.