Overview of the SAS IT Resource Management Data Model :: SAS(R) IT Resource Management 3.3: Administrator's Guide

About the SAS IT Resource Management Data Model

The data model that is supplied for those adapters that are supported by SAS IT Resource Management provides a rich set of measures. The data model is fully documented on the product documentation page: http://support.sas.com/documentation/onlinedoc/itsv/zipfiles/HTMLDoc_3.3/DataModel.html. At this site, the SAS IT Resource Management data model can be interactively navigated, including all of the stage and aggregation (domain category) tables. For each staged table, data and computed columns along with the attributes of each can be reviewed. For each aggregation table, the staged data from which the aggregation was created; the properties of the aggregation; and the attributes of each of the class, ID, statistic, rank, and computed columns can be reviewed.

Reports that are built with this data model are supplied with the solution. Samples of reports are located here: http://support.sas.com/documentation/onlinedoc/itsv/zipfiles/HTMLDoc/Reports.html.

Tip

When designing a user-written adapter, you should review and follow the column labeling scheme of the supplied data model as closely as possible. This practice enables you to copy and reuse those reports in support of your user-written data source. For more information about this topic, see User-Written Staging Transformations.

The data model supports a number of data sources from computer system hardware, operating system software, networks, Web servers, databases, and applications on z/OS, Windows, and UNIX platforms. Distinct data sources are characterized as adapters in SAS IT Resource Management terminology.

Features of the SAS IT Resource Management Data Model

The following characteristics are available from the SAS IT Resource Management data model for all of the supported adapters:

Column labeling is standardized to make the data model meaningful and consistent across adapters in SAS IT Resource Management objects, including template tables, staged tables, aggregation tables, and information maps. Labels are coded for easier viewing and to shorten total label length, making them easier to understand. In most cases, columns with similar meaning in different adapters are named with a similar label in order to promote more consistency in classification and metric names across adapters. In most labels, the name incorporates the respective domain category.
A significant number of computed columns are added to the data model. Computations for computed columns are often simply represented as an RVALUE expression. However, there are some computed columns that include multiple SAS statements in the expression. In addition to these supplied computed columns, users can extend the function of the data model by adding their own computed columns.
Supplied computed columns include the following areas of functionality:
- standard date and time derivations that are based on the source datetime found in the raw data.
- normalized columns that convert lowest common denominator units into industry standard data representations. These normalizations include the following conversions:
  
  bytes to kilobytes, megabytes, and gigabytes
  
  seconds to milliseconds
  
  bits to megabits
  
  service units to MSU (millions of service units)
  
  percentages between 0 and 1 to percentages between 0 and 100
- generation of a total value from several disparate parts—for example, input + output, read + write, received + sent, and so on.
- generation of a column that contains a value that is the opposite of an already existing column. This conversion highlights the variance in the usage of a resource in contrast to the availability of the same resource.
- generation of event counters to count individual events and enable them to be easily summarized when aggregating the data.
- generation of new classification columns when an individual instance of a performance metric needs to be separated from the global instance. This enables the global and individual instances to be aggregated separately.
- conversion of raw counts to rate-based values.
- calculation of buckets for counts and percentages that are associated with ranges of response time.
Formulas are also available in the data model. In SAS IT Resource Management, formulas are reusable shared expressions for use with computed columns. Supplied SAS IT Resource Management formulas are frequently used to set the values for date-based and time-based computed columns that are included in the data model.
The most suitable national language support (NLS) format is used for many supplied columns in the data model. Therefore, a large percentage of columns is ready for formatting in multiple languages. However, some time-based columns that focus on accumulations of time and other columns already using specialized formats have not been converted to NLS. The reason for this is that there is currently not an acceptable equivalent NLS format.

Data Model Staging and Aggregation General Concepts

The process of staging data is required in order to prepare raw data and to create data extracts in SAS data set format. Staging is an intermediate step that is performed before the creation of aggregations in the SAS IT Resource Management data model. (Existing staged data is overwritten for each new iteration of the staging process.) The SAS IT Resource Management data model is based on aggregations. Aggregations are created to provide report-ready data that can be easily input into an information map. An information map is a map that enables the aggregated data to be available for several SAS reporting clients, including SAS Enterprise Guide, SAS Web Report Studio, and SAS OLAP Cube Studio.

Note: The following information describes the aggregations that are generated by means of the Adapter Setup wizard. The information does not necessarily apply to all aggregations in general.

Aggregated data tables in the SAS IT Resource Management data model have the following structures and properties associated with them:

The name of the aggregation table consists of a concatenation of a time period and a descriptive string. The descriptive string represents a domain category in a mixed-case format that uses uppercase letters to signify the beginning of a new word. For example, the name of an aggregation table might be one of the following: DayHourCpu or MonthJobSummary.
The aggregation table has an aging limit specified in days.
The time period for aggregated data is represented by one or more columns in the classification list.
The aggregated data contains one or more classification columns. Some of these columns are time-based and other columns are not based on time.
The aggregated data might contain one or more ID columns.
The aggregated data contains one or more statistics. Statistics can be weighted or unweighted, depending on the nature of the performance metric. Weighting is typically by duration of time or by a counter that indicates a number of events. Weighting columns are explicitly specified in SAS IT Resource Management.
The aggregated data can contain one or more ranked metrics. Each ranked metric can be based on a class or ID variable or on a statistic.
The aggregated data can contain one or more computed columns. These columns can use any aggregated data as sources for their calculations.
Filters can be used both to keep only appropriate data, and to reduce the volume of the aggregation table's output data.

Data Model Aggregation Table Groups and Time Periods

Aggregation tables in the data model are grouped into aggregation table groups. Multiple aggregation table groups can be defined within an adapter's domain category. An aggregation table group is a set of aggregation tables that contain a set of classification columns that are identical, except for the time period classification columns. The time periods vary among the aggregation tables in the aggregation table group.

For example, in an aggregation table group that is based on Memory, the aggregation tables are named DayMemory, DayHourMemory, DayShiftMemory, MonthMemory, MonthHourMemory, MonthShiftMemory, and XMinMemory. (XMinMemory signifies aggregation tables that are based on sub-hourly memory activity). Typical time periods that can be represented in an aggregation table are datetime, hour, shift, day, week, and month. An aggregation table can focus on a single time period, such as datetime, day, week, or month. Alternatively, it can include multiple time periods such as day and hour or day and shift. An aggregation table that uses the datetime period typically focuses on time intervals that are less than a full hour. Aggregation tables that focus on day, week, month, hour, and shift time periods work with time periods of an hour or more.

Data Model Aggregation Key and Ranked Metrics

The SAS IT Resource Management data model provides metrics that are identified as key or ranked metrics. Key and ranked metrics are primarily the same metrics within the adapters, but they are handled differently depending on the type of aggregation that they are used in.

Key metrics are metrics that are output to a key metrics aggregation table. A key metrics aggregation table usually includes the term “KeyMetrics” in its name (for example, KeyMetricsMemory). The intent of the key metrics aggregation table is to keep a limited set of important performance metrics for a long period of time. This tactic facilitates better capacity planning and forecast reporting. Wherever appropriate, key metrics are standardized across adapters. Key metrics are not ranked. Only one statistic is specified for a key metric in a key metrics aggregation table. For specific information about the key and ranked metrics for an adapter and domain category, see What Are Key Metrics?.

Note: For some adapter and domain combinations, such as the SMF adapter and the RMF, CICS, DB2, Jobs, TSO, and OMVS domain categories, the key metrics also provide the ability to filter on the top or bottom rated resource values within a set of classification values. The latter ranking filters are very useful to limit the number of reports that are created to those classifications with the greatest or the least resource utilization.
Ranked metrics are metrics within an aggregation table for which ranking is done. Metrics can be ranked in any aggregation tables. Ranked metrics are based on statistics. Typically, only the most meaningful statistic is ranked.
Both metrics and date classifications can be ranked using either ascending or descending criteria that is based on a list of classifications that are specified for the ranking of the available data.

There are several distinct types of aggregation processing strategies that are available for the supported adapters and domain categories in the data model. Each SAS IT Resource Management supplied adapter can have one or more domain categories that are associated with it. Domain categories have been created that enable users to group source data that should have data staged and aggregated together for reporting purposes.