The primary index tab is only available when connected to a Teradata database. The primary index tab below is available for most analyses that produce an output table containing user data (or more specifically, for analyses in the Reorganization and ADS categories other than Denorm and Refresh). The specific options are explained in the sections defining the output for specific analyses. The default selection is most typically determined by the primary index columns of the input table.
The options displayed on the primary index tab are dependent on the options selected on the storage tab and the release of Teradata in use. In particular, no options are displayed on the primary index tab when creating a database view or explaining the query plan. Note, however, that options are displayed even when the output style is to select data and not create an output table. In this case, the primary index information is used if the analysis is referenced and changed to produce a volatile output table. The Additional Information text box can be used to specify a request for index partitioning , such as PARTITION BY (RANGE_N (cust_id BETWEEN 1362480 AND 1363495 EACH 100)), or in some cases, column partitioning. For more information about column partitioning, see Teradata Columnar Support (Teradata Database).
The option to Create Table with NO PRIMARY INDEX is displayed only when the release of Teradata in use is 13.00 or greater, and a table is being created or results are being selected. When this option is selected, several of the fields are not enabled, as seen in the picture below. Further, if this option is selected, the options relating to collecting statistics on the post processing tab described in Post Processing are not given.
Finally, if the option to Build a Primary Time Index (PTI) is selected, the Additional Information window is replaced with Primary Time Index Fields as shown below.
-
Timecode Data Type
- DATE
- TIMESTAMP(0-6)
- TIMESTAMP(0-6) WITH TIME ZONE
For example: TIMESTAMP(6)
-
Timezero Date
A date marking the earliest data accepted by the time series table. By default, this is DATE ‘1970-01-01’. The sub-fields for this option should have the number of digits indicated by each label. For example, the default value could be entered as YYYY—1970, MM—01, DD—01.
-
Timebucket Duration
- CAL_YEARS(n)
- CAL_MONTHS(n)
- CAL_DAYS(n)
- WEEKS(n)
- DAYS(n)
- HOURS(n)
- MINUTES(n)
- SECONDS(n)
- MILLISECONDS(n)
- MICROSECONDS(n)
Where n is a positive integer indicating the number of time periods.
-
Sequenced
- NONSEQUENCED
- SEQUENCED
- SEQUENCED(max)
Where max is a positive integer specifying the maximum number of rows allowed with the same timestamp.