Data Handling Differences between SAS IT Resource Management 2.6 or 2.7 and 3.3

Table of Data Handling Differences
Function
SAS IT Resource Management 2.6 or 2.7
SAS IT Resource Management 3.3
Storing numeric columns
Note: Numeric values in SAS are stored in floating point, using a maximum of 8 bytes.
Some numeric values are designated to use less than 8 bytes of storage.
%RMPDB2DM unilaterally creates all numeric columns with a length of 8 bytes.
UPDATENUMERICLENGTHS=Y | N. This parameter specifies whether numeric columns keep their length as specified in the PDB.
Aging incoming data
All incoming data is permitted in the DETAIL level, regardless of age limit. Subsequent %CxPROCES invocations aged this data out. This action permitted the backloading of data among other uses.
As %RMPDB2DM copies data from the PDB into the IT data mart, the age limit for the DETAIL label is applied against the data. Thus, it is possible that there are fewer observations in the target detail table than were in the source detail table.
Handling normalized values
The use of variable interpretation types influences the behavior of summarization. Specifically, columns with an interpretation type of COUNT, TIME, or TIMETICKS in INTERVAL tables underwent a normalization transformation before summarization.
In this normalization process, the incoming value is first turned into a rate by dividing the metric by DURATION. If a weight variable has been specified, then the metric would instead be divided by the weight variable. The resulting value is the value that was used in computations by %CPREDUCE.
%RMPDB2DM creates a corresponding rate column in the staged table for each COUNT, TIME, and TIMETICKS type column migrated for an interval table. This is done to provide results that are consistent with the existing data that migrated from SAS IT Resource Management 2.6 or 2.7. This rate column is then used with weighted statistics (such as weighted mean), specifying DURATION or the indicated SAS IT Resource Management 2.6 or 2.7 weight variable. This provides the same computations that are achieved through the normalization technique used in SAS IT Resource Management 2.6 or 2.7 .
For example, if you had a COUNT column named PGFAULT in an SAS IT Resource Management 2.6 or 2.7 interval table, then the SAS IT Resource Management 3.3 staged table would include a computed column named PGFAULT_RATE. The source code for the computation would be as follows:
IF DURATION GT 0
 THEN PGFAULT_RATE = PGFAULT / DURATION;
ELSE PGFAULT_RATE = .;
Handling pre-set values for SHIFT
It is possible to set the value for SHIFT in the underlying data (for example, through MXG customization), and not use the SAS IT Resource Management SHIFT definitions.
This capability is no longer supported. SHIFT is a computed column in the tables that are produced by staging transformations. This computed column uses a MIGRATION_SHIFT formula that is created by %RMPDB2DM.
If you want to alter the logic for assigning SHIFT, then you can make either of these modifications:
  • Change the source code for the MIGRATION_SHIFT formula. This change affects the computation of SHIFT in all staged tables in the IT data mart. You need to locate the MIGRATION_SHIFT formula in the PDB Migration Formulas folder of the IT data mart for this invocation of %RMPDB2DM.
  • Change the source code for the SHIFT computed column in the desired staged tables. This change affects only this particular SHIFT column.
For information about how to handle holidays, see “Handling Holidays” in the “Best Practices Appendix”of the SAS IT Resource Management: Administrator’s Guide.
Supporting HP-PCS tables
Support is provided for the following tables for the HP-PCS collector:
  • PCSGDK (PCS Global Disk Metrics)
  • PCSGLN (PCS Global LAN Metrics)
In SAS IT Resource Management 2.6 and 2.7, these tables were populated by retrieving data from the PCSGLB (Global) table.
Support is not provided for the following tables for the HP-PCS collector with a staging transformation:
  • PCSGDK (PCS Global Disk Metrics)
  • PCSGLN (PCS Global LAN Metrics)
In SAS IT Resource Management 3.3, use the PCSGLB table instead. The PCSGDK and PCSGLN staged tables and aggregations are migrated. However, the staging transformation does not populate these tables.
Handling Web Log data with respect to migration
The Web Log adapter works by presummarizing data to an hourly granularity into the WEBRES or WEBCLI table. It then used the DETAIL parameter of %CPREDUCE() to populate the required tables.
The WEBLOG staging transformation does not support the WEBRES or WEBCLI tables. Instead, the staging transformation directly populates the required staged tables.
Thus, for PDB migration, the staged and aggregation tables are created for all Web Log tables that are marked KEPT=’Y’ in the PDB, with the exception of WEBCLI and WEBRES.
Supporting Sum of Weights statistic
Note: Both SAS IT Resource Management 2.6 or 2.7 and SAS IT Resource Management 3.3 enable you to request computation of weighted statistics. This request typically involves the computation of the sum of weights.
The sum of weights is treated as the sum of the weight variable. For example, if X (the analysis variable) is to be weighted by Y, then the sum of weights is simply the sum statistic requested on Y.
In cases where the analysis has no missing values, this treatment is entirely correct. However, if the analysis variable has a missing value, then this treatment can lead to erroneous results.
SAS IT Resource Management 3.3 correctly maintains the sum of weights for an analysis variable as a separate statistic. This statistic is the sum of the weight variable where the analysis variable is not missing. Thus, it is the sum of Y with respect to X not missing.