Baselining is a process by which Traverse can automatically set the warning and critical thresholds for each test based on the test's historical data. This allows one to set customized thresholds automatically based on each tests's individual behavior.
As an example, the response time for a local device is normally much smaller than the response time for a device in a remote datacenter because of network latency. Rather than setting the response time warning threshold for all devices to be the same, you can use the baseline feature to calculate the 95th percentile of the response time reported for each device over a three-month period, and then set the warning threshold to be 10% higher than this 95th percentile value.
Once a baseline threshold value is set for a test, the threshold value is static. If you wish to re-calibrate the baseline threshold, you need to rerun it.
Baseline Data Set
The baseline value is calculated for each test based on its own historical data. You select the devices and tests for which you want to run baselining by specifying a combination of device name, test name and test type.
Each time Traverse aggregates a test result, it stores three values: The minimum, maximum, and mean values of the tested variable over the course of the aggregation period. For example, if Traverse is configured to store data for 1 day at 10 minute samples, and a test is set up to run every 10 minutes, in the course of a day it generates 144 test results. Each test result includes the maximum, minimum, and mean values of the tested quantity for the 10 minute period. You can generate a baseline from the maximum, minimum, or mean samples within the specified date range.
Traverse can calculate a single baseline value based on the historical data which can then be used to generate a static warning and critical threshold for a test. In addition to static thresholds, Traverse can also calculate the baseline per day of week and per hour of day (e.g. 8am on Thu) and use these dynamic baselines to create time based thresholds.
Creating a Baseline and Setting Thresholds for One or More Tests
Field |
Description |
Device Name/RegExp |
The name of a device whose tests are to be baselined, or a regular expression containing `*' wildcards to match multiple device names. |
TestName/RegExp |
The name of an individual test to be baselined, or a regular expression containing the `*' wildcards to match multiple test names. |
Test Type/Subtype |
The monitor and subtype of the test(s) to be baselined. e.g. port/http, snmp/chassis_temp. |
Start Date, End Date |
The start and end date of the test results to be used in calculating the baseline. Note: Each selected test must have test results available for the full date range. |
Taking values of |
The value from each test result (maximum, minimium, or mean) that is used to calculate the baseline. |
And using the |
The method (average or 95th percentile) used to calculate the baseline from the maximum, minimum, or mean test results. average is the mean of the test results (sum of test results / number of test results). |
Warning Threshold |
A percentage above or below the calculated baseline. Select above if the test result gets worse as it gets higher. Select below if the test result gets worse as it gets lower. When the test result crosses this threshold, test status is set to Warning. |
Critical Threshold |
A percentage above or below the calculated baseline. Select above if the test result gets worse as it gets higher. Select below if the test result gets worse as it gets lower. When the test result crosses this threshold, test status is set to Critical. |