Next Topic

Previous Topic

Book Contents

Counter Thresholds

Monitor > Monitor Sets

  • Select a monitor set in a folder, then Counter Thresholds

The Counter Thresholds tab defines alarm conditions for all performance objects/instances/counters associated with a monitor set. These are the same performance objects, instances and counters displayed when you run PerfMon.exe on a Windows machine.

Performance Objects, Instances and Counters

When setting up counter thresholds in monitor sets, it's helpful to keep in mind exactly how both Windows and the VSA identify the components you can monitor:

  • Performance Object - A logical collection of counters that is associated with a resource or service that can be monitored. For example: processors, memory, physical disks, servers each have their own sets of predefined counters.
  • Performance Object Instance - A term used to distinguish between multiple performance objects of the same type on a computer. For example: multiple processors or multiple physical disks. The VSA lets you skip this field if there is only one instance of an object.
  • Performance Counter - A data item that is associated with a performance object, and if necessary, the instance. Each selected counter presents a value corresponding to a particular aspect of the performance that is defined for the performance object and instance.

Select Page

When more rows of data are selected than can be displayed on a single page, click the and buttons to display the previous and next page. The drop-down list alphabetically lists the first record of each page of data.

Edit icon

Click the edit icon next to row to edit the row.

Delete Icon

Click the delete icon to delete this record.

Add / Edit

Click Add or the edit icon to use a wizard that leads you through the six steps required to add or edit a performance counter.

  1. Select a Object, Counter and, if necessary, an Instance using their respective drop-down lists.
    • If only one instance of a performance object exists, the Instance field can usually be skipped.
    • The drop-down lists used to select performance objects, counters, and instances are based on the "master list" maintained using the Monitor Lists page. If an object/instance/counter does not display in its respective drop-down list, you can add it manually using Add Object, Add Counter, and Add Instance. You can also update the "master list" of all objects, instances and counters by scanning specific machine IDs using Update Lists By Scan. Once the update is completed, the drop lists should be populated with the options you require.
    • When multiple instances exist, you often have the option of using an instance called _Total. The _Total instance means you want to monitor the combined value of all the other instances of a performance object as a single counter. The _Total can be used as a kind of "wildcard instance". Without the _Total instance you would have to specify each instance by its exact name, which makes applying the same monitor set to multiple machines difficult. The true benefit of the _Total instance is determining if there are any performance issues for any instance of this object at all. Once you know that you can investigate the specific cause.
    • When multiple instances exist, you sometimes have the option of using an instance called *ALL. The *ALL instance means you want to monitor all instances for the same performance object using individual counters.
  2. Optionally change the default counter object Name and Description.
  3. Select the log data collected. If the returned value is numeric, you can minimize unwanted log data by setting a collection operator just over or just under the collection threshold.
    • Collection Operator - For character string return values, the options are Changed, Equal or NotEqual. For numeric return values, the options are Equal, NotEqual, Over, or Under.
    • Collection Threshold - Set a fixed value that the returned value is compared to, using the selected Collection Operator, to determine what log data is collected.
    • Sample Interval - Defines how frequently the data is sent by the agent to the KServer.
  4. Specify when an alarm condition is encountered.
    • Alarm Operator - For character string return values, the options are Changed, Equal or NotEqual. For numeric return values, the options are Equal, NotEqual, Over or Under.
    • Alarm Threshold - Set a fixed value that the returned value is compared to, using the selected Alarm Operator, to determine when an alarm condition is encountered.
    • Duration - Specify the time the returned values must continuously exceed the alarm threshold to generate the alarm condition. Many alarm conditions are only alarming if the level is sustained over a long period of time.
    • Ignore additional alarms for - Suppress additional alarm conditions for this same issue for this time period. This reduces the confusion of many alarm conditions for the same issue.
  5. Warn when within X% of alarm threshold - Optionally display a warning alarm condition when the returned value is within a specified percentage of the Alarm Threshold. The default warning icon is a yellow traffic light icon . See Monitor Icons.
  6. Optionally activate a trending alarm. Trending alarms use historical data to predict when the next alarm condition will occur.
    • Trending Activated? - If yes, a linear regression trendline is calculated based on the last 2500 data points logged.
    • Trending Window - The time period used to extend the calculated trendline into the future. If the predicted trendline exceeds the alarm threshold within the future time period specified, a trending alarm condition is generated. Typically a trending window should be set to the amount of time you need to prepare for an alarm condition, if it occurs. Example: a user may want 10 days notice before a hard drive reaches the alarm condition, to accommodate ordering, shipping and installing a larger hard drive.
    • Ignore additional trending alarms for - Suppress additional trending alarm conditions for this same issue for this time period.
    • By default, trending alarms display as an orange icon . You can change this icon using the Monitor Icons tab.

    Warning status alarm conditions and trending status alarm conditions don't create alarm entries in the alarm log, but they change the image of the alarm icon in various display windows. You can generate a trending alarm report using Reports > Monitor.

Next

Moves to the next wizard page.

Previous

Moves back to the previous wizard page.

Save

Saves changes to a record.

Cancel

Ignores changes and returns to the list of records.