Working with custom statistics
Use the Custom statistics page in the TrueSight Capacity Optimization console to analyze and manage the custom statistics generated by the warehousing engine. Custom statistics can be rolling, produced-based, or rule-based.
The Custom statistics page displays tabs for each type, allowing you to overview producer-based statistics and manage rule-based and rolling statistics. To access the Custom statistics page, click Administration > DATA WAREHOUSE > Custom statistics.
For more information on types of custom statistics, see the following:
Rolling statistics
Rolling statistics is a combination of entity type and metric, and created by computing percentile statistics on a metric. These percentile values are computed for the last 1,7, and 30 days for the selected metrics.
TrueSight Capacity Optimization includes the rolling statistics for predefined set of entities and metrics. Right-click here and select Save link as to download the Excel file that includes the list of out-of-the-box rolling statistics.
If you want to compute rolling statistics for additional metrics, use the Rolling Statistics tab to create custom rolling statistics. This tab displays a list of the existing out-of-the-box and custom (user-defined) rolling statistics and enables you to define new ones.
By default, the custom views created using the Tree map and Sunburst templates use these rolling statistics.
This tab is available only when you apply Service Pack 1 (11.5.01) of TrueSight Capacity Optimization 11.5.
The Rolling statistics table displays the following:
- System/business driver type: Entity type associated with a metric of the custom statistics.
- Metric name: Name of the metric.
- Package: Package for the statistics.
- Extension for Distributed Systems: Indicates that this rolling statistic is an out-of-the-box statistic.
- [Custom]: Indicates that this rolling statistic is a user-defined statistic.
- Type: Entity category of the statistic.
To add rolling statistics
- Navigate to Administration > Data Warehouse > Custom statistics. By default, the Rolling Statistics tab is displayed.
In the Rolling statistics tab, select Actions, and click Add Rolling statistics.
In the Add Rolling statistics page, specify the following details:
- Entity category: Select an entity category as System or Business Driver.
- Entity type: Select entity types based on the selected entity category.
- Click Edit.
In Available items, a list of entity types corresponding to the selected entity category is displayed. For example, if you have selected System as an entity category, all available system types are displayed.
Select the entity type and click to move to Selected Items.
Click Apply. The selected entity types are displayed.
- Metric: Select metrics based on the selected entity types.
- Click Edit.
- In Available items, several categories of metrics corresponding to the selected entity types are displayed.
- Select the metrics and click to move to Selected Items.
- Click Apply. The selected metrics are displayed.
- Click Save.
The rolling statistics are calculated based on last 1, 7, and 30 days of data.
Producer-based statistics
Producer-based statistics are generated by an object called Producer, which can be:
- A model that generates predictions (for example, the residual capacity of an application)
- An analysis that summarizes collected data (for example, the weekly trend of a system KPI)
Not all models or analyses can produce custom statistics. When you run analyses and models in the console, the time series data that is produced is not included in the public views. You can view the charts and download the underlying data, but the data is not saved in the database. However, a user with administrative privileges can promote any analysis or model to golden and the analysis or model is marked as a golden analysis or model. The Golden producer tasks (background tasks) run the golden analyses and models automatically on a periodic basis, and save the produced data in the database as a “producer-based” time series.
Tip
Golden models and analyses store their results in a permanent way and can produce custom statistics.
To save a producer-based statistic
- Create a model or analysis as described in Editing and maintaining models or Creating an analysis
- Run the model or analysis and check the correctness of its results
- Using an administration account, visit the main page of the new model or analysis and click Promote to Golden.
The Promote to Golden option - Select the publishing period, that is the update frequency and click Save; the available options are: daily, weekly or monthly
Promoting an analysis to Golden and specifying the publishing period
The production of custom statistics is now enabled. A scheduler task runs the golden model or analysis with the specified frequency, and the produced results are displayed in the Custom stats page.
Rule-based statistics
Rule-based statistics are calculated by the data warehouse according to administrator-specified rules. You can add, edit, or delete these stats using the icons provided in the table. Only rule-based statistics that have no defined metric associations can be deleted.
To create a new rule-based statistic, define the statistic, and then add system and business driver metrics to it.
To create a new rule-based statistic
- Navigate to Administration > Data Warehouse > Custom statistics.
In the Rule-based statistics table, click Add stat.
In the Add new rule-based statistic page, specify the following details:
Property Value Parameters
Name Enter the name of the new rule-based statistic. Description Enter a brief description for this rule-based statistic. Type The type of statistic you intend to use. Select any one: - Advanced aggregation function: You can aggregate your data using aggregation functions.
- Percentile: You can calculate the percentile.
Advanced aggregation function Input resolution Select the input resolution. For example, if you want to create a statistic based on hourly values, you can decide to use Hour. Select any one: Detail, Hour, Day, or Month. Input time period Select any one:
- time filter: Lets you define the period over which the statistic will be calculated by selecting a time span from the Time filter list.
- Select the Time filter.
- manually specified: Lets you define a specific time span by allowing you to enter a particular time frame. This time frame can be specified in hours, days, weeks, months, years by selecting from the list displayed for Manually specified. You can also include the current time frame by selecting the checkbox next to the list. The name of this check box changes depending on the option you select in the list. For example, if you select days in the list, the check box name displayed is Including current day.
- Specify the details like in Years, Months, Weeks, Days, or Hours w/o Including current hour.
Aggregation Allows you to select which kind of aggregation you want to apply to the input data. For example Weighted average. The available options are Average, Weighted average, Duration weighted average, Minimum, Maximum, Sum, Expected sum, or Count. Output The time resolution of the output. The available options are A single point, A point for each hour, or A point for each hour class. Percentile Input resolution Lets you select the input data. For example, if you want to create a statistic based on hourly values, you can decide to use Hour. Select any one: Raw, Detail, Hour, or Day. Time aggregation The aggregation of the input data on which you want to calculate the statistic. Select Daily, Monthly, or Quarterly as appropriate. Percentile order (P%) Enter a percentage value to specify the distribution of data. Note
To view or configure Advanced properties, click Advanced. You do not need to set or modify these properties unless you want to change the way the rule-based statistic works. These properties are for advanced users and scenarios only.
Property Value Parameters
Advanced aggregation function Input statistic Select any one: Automatic - based on metric type, Average, Weighted average, Minimum, Maximum, Sum, Quadratic sum, Cumulative sum, Good samples, Warning samples, or Poor samples.
Aggregation groups Allows you to group the input data depending on the aggregate condition described as follows: - none: Select this option if you do not want to aggregate the result achieved after applying this rule.
- aggregate points having: Select this option to specify one of the following aggregate conditions for applying the rule.
- Select any of the Having conditions options: The same day class, The same week day, and The same month day.
- Select any of the Having conditions options: The same day class, The same week day, and The same month day.
Percentile Type of percentile Select either Discrete (to pick data in its current form) or Continuous (for uniform distribution of data) to specify how data must be configured. Input statistic Select any one: Automatic- based on metric type, Average, Weighted average, Minimum, Maximum, Sum, Quadraticsum, or Cumulativesum. Order Lets you specify how the data used to calculate the statistic should be sorted (Ascending or Descending). - Click Save to create the statistic.
To add metrics to the statistic
- Click the statistic's name in the Rule-based statistics table to open its detail page.
- Click Add system metric, Add business driver metric, or Add global metric.
- Add metrics and click Save.
Detail page for a custom rule-based statistic
As you create new associations, they will be listed in the corresponding table of the statistic's detail page.
The Historical recovery icon enables you to recover historical data, while the Delete the association with this custom statistic icon lets you delete the association.
Comments
Log in or register to comment.