Creating a historical collection

A historical data collection specifies the attribute group to collect data from, where to store the historical data, and other information such as the collection frequency and distribution. Create a historical collection definition for every attribute group that you want to collect historical data for. You can then retrieve the historical data into query-based views.

1. Click History Configuration to open the History Collection Configuration window.
2. Click Create a new collection.
   If you first click a Monitored Application (or right-click and click ), Monitored Application will be selected for you.
3. Enter a Name of up to 256 bytes. A short name is also given to the collection and is shown in the middle section of the status bar.
4. Optional: Enter a Description for the collection, up to 64 bytes.
5. Select a Monitored Application from the list.
6. Select an Attribute Group from the list.
   Not all the product attribute groups necessarily display: only those that are appropriate for historical collection and reporting.
7. Click OK to open the configuration tabs for the collection.
   The branch of the monitored applications expands to show the new collection. At this point, it has not been distributed to the managed systems.
8. Complete the fields in the Basic tab:
   1. Collection Interval is the frequency of data transmission to the short-term history file on the computer where the data is saved (Cloud Pak System Software Monitoring Agent or Cloud Pak System Software Monitoring Server). The options are every 1, 5, 15, or 30 minutes, every hour, or once per day. The default interval is 15 minutes.
      
      The shorter the interval, the faster and larger the history file grows. This can overload the database, warehouse proxy, and summarization and pruning agent. For example, if you set a 1-minute collection interval for Process data, expect the summarization and pruning for that attribute group to take a long time. Such a short interval should be enabled for an attribute group only if it is critical in your work.
   2. Collection Location is where the short-term historical data file resides: at the TEMA (Tivoli Enterprise Monitoring Agent) or the TEMS (Cloud Pak System Software Monitoring Server). The default location is TEMA, which minimizes the performance impact on the monitoring server from historical data management.
      
      If you are filtering the historical collection (step 9), collection at the TEMA also gives you more space for composing the filter. The formula example at the end of this topic takes up 63% of the formula storage when collection is set to TEMS; and 31% when collection is set to TEMA..
      Note, however, that TEMS might be a better choice for certain environments. Also, the OMEGAMON XE on z/OS product requires that the data be stored at the monitoring server.
   3. Warehouse Interval determines whether the collected data is warehoused and how often. The options are 15 minutes, 30 minutes, 1 hour, 12 hours, 1 day, or Off.
      
      A more frequent warehousing interval enables quicker availability of warehoused data for retrieval. Shorter intervals cause some additional processing to check for and transmit newly collected data; and there are more frequent elevated levels of transmission activity, but for shorter durations. Regardless of the warehouse interval, the most recent 24 hours are always available in the short-term history files. If you use Tivoli Common Reporting, note that queries are to the data warehouse only.
9. If you want to restrict the collection of samples to only those that meet certain conditions, click the Filter tab and compose the formula:
   1. Click Add Attributes.
   2. Select the attribute to filter on and click OK. You can use Ctrl + click to select multiple attributes, Shift + click to select all entries between the attribute previously selected and this one, or Select All to select every attribute.
   3. For each attribute selected, click inside the cell under the column heading and compose an expression consisting of a function, a relational operator and a test value:
      To change the function, click Value (or Compare date and time for a time attribute) and select one from the list. See Formula Functions for the syntax, description, and an example of each function.
      To change the relational operator, click equal and select another operator from the list.
      Click inside the text box and enter or select the value.
   4. Keep multiple expressions in the same row if they must all be met (Boolean AND logic) and on separate rows if any of them might be met (Boolean OR logic) for the sample to be saved in the short-term history file.
   5. Select the same attribute twice if you want to collect a range, such as > 1 AND < 90. Use Add Attributes to select the attribute again. (See the second example at the end of this topic.)
   6. Confirm that the Formula capacity counter is below 100%. Setting the Collection Location (step 8) to TEMA gives the most space.
10. Click Apply to save the collection.

Results

The collection name appears in the monitored application branch of the tree with an icon to indicate that it is not distributed to managed systems: .

Example

This is a historical collection called ProcessInformation. The settings in the Basic tab use the defaults: Collection Interval of 5 minutes; Collection Location at the TEMA; and Warehouse Interval of 15 minutes. The Filter tab has a formula written to collect the sample if one of these two conditions is met: the process name is VirusScan; OR the process name is java and is taking more than 50% of the processor time on the North012 Windows system, the South007 Windows system, or both.

	Process Name	% Processor Time	Server Name
1	==VirusScan
2	==java	>50	('Primary:North012:NT', 'Primary:South007:NT')

The IN operator was selected for entering the server list, as you can see when viewing the formula in :

(Process Name==VirusScan) OR (Process Name==java AND %Processor Time >50
AND Server Name *IN ('Primary:North012:NT','Primary:South007:NT'))

The following example shows a filter formula with a repeating attribute:

	% Processor Time	% Processor Time
1	>40	<60