Baseline Thresholds

By default the baseline threshold calculator uses the last seven days of collected statistic values to determine what is "normal" using mathematical standard deviation calculations. You could dump the collected statistical information into Excel and run similar formulas yourself, but the Baseline Calculator simplifies this process for you.

Is this a correct way to set the baseline thresholds of 7 days (default setting, understand that it can be changed)?       - A new template is created and NOT to  set any threshold in the template.      - Apply this new template to a node or nodes, wait for 7 days then set threshold at the “Application” level by select the option to “Override Template” and “Use Latest Baseline  Thresholds”

Yes, though it is recommended that if you have a contextual understanding of the metric you are monitoring that you define these thresholds yourself manually. E.G. Percent Disk Space usage can probably be defined on your own without much assistance from the baseline calculator. You can also configure the template to use baseline thresholds, which after seven days of collection will automatically calculate and apply thresholds (where possible) to the component without any manual effort.

After the initial 7 days setting of the threshold applied, in what scenarios should the thresholds be recalculated\reset again?

By definition a baseline is a period when things are operating normal within the environment. Any anomalies which occur during the baseline period will be calculated into the results and skew the recommended values. If this occurs you should re-baseline to ensure the recommended values are as accurate as possible. Another reason why you might re-baseline is if there are significant changes made which which would influence what "normal" is for the environment. This could include merging with or acquiring a new company and onboarding a large number of new users to an existing application, or making substantive improvements to the supporting infrastructure.

Baseline Thresholds error “Thresholds could not be calculated, standard deviation is zero”, what can it be done to set the thresholds correctly? Manual setting?

This error occurs when there has been no statistically significant deviation of the collected metric during the baseline period. Without some variance in the integer value polled, there is no way of determining a valid warning or critical threshold. E.G. If I poll the Disk I/O stats on a dormant volume for seven days and the value is always zero, there is no way statistically to know if "1" is bad or if "1,000,000" is still good.

Baseline Thresholds Warning “The recommended warning and/or critical thresholds are less then minimum value for this statistic…”, what can it be done to set the thresholds correctly? Manual setting?

This error usually occurs when the Less Than, Greater Than operators are changed but the values are not updated. For instance when you have a "Greater Than" operator your Warning Threshold cannot be greater than your "Critical" Threshold. The inverse is true when the "Less Than" operator is used.

Hi,

These values are calculated once a day and are set for the whole day until the next recalculation?

And are the same for night hours and business hours?

Hi,

These values are calculated once a day and are set for the whole day until the next recalculation?

And are the same for night hours and business hours?

The definition of a baseline is a period of time when things are deemed "normal". As such, the baseline period is determined by you using the last seven days of historical data. Once a baseline has been set it is not recalculated until you force a recalculation to occur. This would typically be when there have been significant changes made in the environment such as additional resources added to the system being monitored, or major software upgrade which may make your previous baseline invalid.

alterego,

Is my interpretation of the following incorrect? It seems to suggest that node baselines are calculated automatically, but that the schedule cannot be changed like it can for interfaces:

What you are referencing above is for node baselines, which function differently than application baselines.

I know this is a long time after the subject, but are you saying that if I choose baselines for CPU/MEM, that they don't populate with a rolling 7 days.  It's calculated once.  If I want new metrics (added capacity, etc), I need to manually run the baselines on my nodes again? aLTeReGo​

It depends. Are you referring to Application or node specific baselines?

Sooo, I know I'm all late but I want to be clear about my understanding on this point.  You are referring to the Node/system resource (CPU/MEM) consumption vs the Application resource (cpu/mem) consumption...right?

Thanks all for this thread.  (sorry I'm late)