17.10 - Using Ferret Parameters - Advanced SQL Engine - Teradata Database

Teradata Vantage™ - Database Utilities

Advanced SQL Engine
Teradata Database
Release Number
Release Date
July 2021
Content Type
Publication ID
English (United States)

The variable, parameter, in the Ferret syntax diagram includes various types of parameters, including multitoken parameters and decimal and hexadecimal numeric input.

Multitoken Parameters

Multitoken parameters, such as subtable identifiers, which are typed as three values, are typed on a single line with one or more spaces separating the individual tokens. You can also enter multitoken parameters separated by hyphens.

For example, either of the following as subtable identifiers is acceptable:

0 200 0

You can separate parameters from each other and from the command by spaces or a comma. To specify a command option (cmdoption ), type it on the same line as the command (cmd).

Numeric Input

Numeric values can be entered in either decimal or hexadecimal format. The default numeric base for data input to and output depends on the radix settings for input and output:

  • When the radix for input is decimal, numeric input is interpreted as decimal numbers. For example, input of 45 is interpreted as the value 45.
  • When the radix for input in hexadecimal, numeric input is interpreted as hexadecimal numbers. For example, input of 45 is interpreted as the value 69.

The initial radix setting is hexadecimal for input and output. For more information on setting the radix, see the documentation for the RADIX command.

The following special numeric formatting conventions can be used to force entered numeric values to be interpreted as decimal or hexadecimal, regardless of the radix setting:

  • Decimal values can be signified by adding a 0i or i prefix, or by adding a decimal point suffix:
    0i45 45. 0I45
  • Hexadecimal values can be signified by adding a 0x or x prefix, or by adding an h suffix. Leading zeros are optional:
    0x2D X2D 2Dh 002DH

The valid range of for numeric values is unsigned 16-bit values, 0 through 65535 (0x0 through 0xFFFF), except when patching and using /L or /W for 32-bit integers, in which case the valid range is 0 through 4,294,967,295 (0x0 through 0xFFFFFFFF).

You must separate two numbers on the same line from each other by a space or a comma.

Specifying a Subtable Identifier (tid)

Commands operate on subtables, rather than on tables. A table is a collection of subtables, and a subtable is a collection of physical rows. Each subtable has physical rows that store one particular type of data. For example, primary and fallback data subtables have data rows, primary and fallback index subtables have index rows, and the physical rows of table header subtables contain table metadata.

Each subtable is identified by a subtable identifier ( tid ). The tid has two parts. The first part identifies the table to which the subtable belongs. All subtables that are part of the same table have the same table identifier as part of their tid. The second part identifies one of the subtables that comprise the table, and indicate the subtable type.

A tid is defined as follows:

{ table_unique | = } type_and_index

table_unique and type_and_index together uniquely identify a subtable.

Syntax Element


table_unique Unique identifier of the table to be acted upon.

The table can be uniquely identified in either of two ways:

  1. Specify the name of the database to which the table belongs and the table name, separated by a period. The names individually or together must be delimited. The following formats are valid:
    • " database_name . table_name"
    • " database_name "." table_name "
    • ' database_name . table_name'
    • ' database_name '.' table_name'
  2. Specify the unique numeric identifier of the table, which consists of two numeric values separated by a space. The table identifier consists of the first two numeric values returned by the TABLEID command. These two numbers are common to all subtables that comprise the table.
    The format of the input numbers depends on the current radix setting, which is displayed by the RADIX command.
= Specifies to use the most recently saved value for table_unique .
type_and_index Uniquely identifies the subtable within the group of subtables that comprise a table. The type_and_index identifies both the type of subtable, and the individual subtable itself.

The syntax for type_and _index is as follows:

{ type [ / index [ / variant ] ] | number }
Syntax Element Description
type The type of subtable.
type can be one of the following:
Type Description
* All the subtables of this table
H The table header subtable.

Because table header subtables have only one row and no secondary indexes or work subtables, any index and variant specifications that follow an H type are ignored.

P The primary data subtable.
F number The fallback data subtable specified by number. The default is 1.
F * All of the fallback subtables.
index The index subtable to examine. If an index subtable value is not provided, it defaults to the data subtable. Index 1 is the first secondary index. index 2 is the second secondary index, and so forth.

index can be one of the following:

index Value Description
number A secondary index that can be used instead of specifying X number . if you enter number by itself, it must be a multiple of 4, and it is interpreted in the following ways:
  • 0 is the same as entering X0
  • 4 is the same as entering X1
  • 8 is the same as entering X2
  • 12 is the same as entering X3
and so forth.
* All the indexes of the table.
D The primary data index (same as X0 or 0).
X number The secondary index specified by number. The default is 1.
X * All the secondary indexes, starting at 0.
variant The possible subtables.
variant can be one of the following:
variant Value Description
* All of the possible variant subtables.
0 The default used if a variant value is not specified.
1 The value during a Sort or Table Modify operation.
2 Unused.
3 Unused.
number A single number that represents internally the type of subtable (header, primary data, or fallback data), the index to use when ordering the physical rows (primary data index or one of the secondary indexes), and the variant. The table header has a type_and_index number of zero. Examples of these numbers are listed below.

The following table gives examples that describe the type_and _index fields.

Subtable Description Number Type Type/Index Type/Index/Variant
Table header 0 H    
Primary data subtable 1024 (0x0400) P P/D  
First secondary index 1028 (0x0404)   P/X1  
First fallback table 2048 (0x0800) F1 F1/D  
Second secondary index of the third fallback table 4104 (0x1008)   F3/X2  
All primary subtables     P/*  
All primary secondary indexes     P/X*  
All fallback subtables   F*    
All subtables of this table   *    
Sort table of the first secondary index 1029 (0x0405)     P/X1/1
Both tables during a sort of the first secondary index       P/X1/ *

For example, assume that table T4 is a table in database MYDB and has a table_number of 0 1198. Also assume that input is accepted in hexadecimal format.

Some valid specifications of a tid for primary subtables of table T4 are as follows:

  • "MYDB.T4" 400
  • "MYDB.T4" 1024.
  • "MYDB"."T4" P
  • u 'MYDB'.'T4' 400 h
  • 0 1198 400
  • 0 1198 P

Classes of Tables

Vantage differentiates among the following five classes of tables. You can specify a table classification when defining the SCOPE parameters of an action.

Table Type Description
Permanent Journal (PJ) Tables Tables that can survive system restarts, but do not contain user-visible data.

Each PJ table contains data generated internally by Vantage. The PJ data is usually used to restore the journaled tables to a given checkpointed state by rolling transactions forward or backwards from an archived copy of the tables.

Permanent Tables Tables containing the real data, which can survive system restarts.
Temporary Tables Tables that can exist as either global or volatile temporary tables as defined below:
  • Global temporary tables exist only during the duration of the SQL session in which they are used.
  • Volatile temporary tables reside in memory and do not survive a system restart. They are treated like spool tables, discussed below.
Spool Tables Tables that contain non-permanent data and can be divided into classes according to their scope of persistence.
Intermediate result spool tables hold temporary results during the processing of a single SQL query and persist only for the duration of that processing. Response spool tables hold the final answer set from a query and a limited number can optionally persist across further queries in the same session. Spool tables can be discarded as follows:
  • Normally, when they are no longer needed.
  • As part of a specific resource cleanup on a transaction abort or session logoff.
  • As part of a general resource cleanup every time the system restarts.
  • Rows for volatile tables are placed in spool spaces and are discarded at the end of a transaction or at the end of a session (depending on a table option or by a DROP TABLE statement).
  • Volatile table definitions reside in memory and do not survive a system restart.
Persist ant Spool Tables These are treated like spool tables, but they survive system restarts.

The attributes associated with each class of tables can affect system performance, since the attributes are set individually, and each class of tables is used for a different purpose.

For example, you might want to pack only Permanent and PJ tables. Therefore, you would specify these tables when defining the SCOPE of the PACKDISK command.

Rows and Row Ranges

Every physical row is uniquely identified by a row ID. A row ID specification, also referred to as a ridspec or a rowspec, is a value that includes a partition number, rowhash or hash bucket number, and uniqueness value. It can occupy up to 16 bytes in a physical row:
  • The partition number is stored as 2 bytes if the number of partitions defined for the table is up to 65535, or 8 bytes if the number is greater. If the table is not partitioned, no space is reserved in the row ID for a partition specification. A value of zero means the table is not partitioned.
  • The specification of the remainder of the row ID depends on whether the table includes a primary index:
    • PI tables have a 4-byte rowhash value and a 4-byte uniqueness value, which is used to impose uniqueness in cases of rowhash synonyms for different physical rows.
    • Tables without a primary index (NoPI tables) and tables with a primary AMP index (PA tables) have a 16- or 20-bit hash bucket value and a 44-bit uniqueness value. For systems using 16-bit hash bucket values, the four bits between the hash bucket value and the uniqueness value are unused.
      The remainder of the RowID is treated as two 4-byte values, and present them as hash and uniq fields:
      hash0 hash1 uniq0 uniq1
      ----- ----- ----- ----- 
      FFA9  CB78  0000  0002

      For NoPI and PA tables displayed or manipulated, hash0 and hash1 actually represent a 20-bit hash bucket value and the 12 high-order bits of the uniqueness value, while uniq0 and uniq1 display the 32 low-order bits of the uniqueness value. For more information on NoPI and PA tables, see Teradata Vantage™ - SQL Data Definition Language Syntax and Examples, B035-1144.

A range of rows, also called a ridrange , is specified by the ridspec of the starting and ending rows:

start_rid_spec [ TO end_rid_spec ]

For purposes of input, the required and optional components of a ridspec depend on whether the table containing the physical row is partitioned.

Table Type Ridspec Syntax
partitioned partitioned_ridspec:
{ row_partition | = } [ { row_hash | = } [ row_unique | = ] ]
nonpartitioned nonpartitioned_ridspec:
{ partitioned_ridspec |
  [ row_partition | = ] { row_hash | = } [ row_unique | = ] 
Syntax Element Description
row_partition The partition number of the physical row. The partition number is a single numeric value. It is the internal partition number used by the file system. The value must be between 0 and 9223372036854775807 (0 and 7FFFFFFFFFFFFFFF in hexadecimal notation).

For tables with multiple levels of partitioning, each physical partition results from the combination of all partitioning expressions in the PARTITION BY clause for a specific set of values in the partitioning columns. This gives a single, unique combined partition number, which is mapped to an internal partition number. Although the combined partition number of a partition can change if partition ranges are dropped or added for the partition level, the internal partition number does not change.

For rows in nonpartitioned tables, the partition number must be zero, if specified, or can be omitted from the ridspec.

row_hash The row hash value of the physical row. The row hash determines the AMP to which the physical row is distributed. The row hash consists of 2 numeric values, usually shown in output screens under column headings hash0 and hash1.
For start_rid_spec :
  • row_hash defaults to 0x0000 0x0000.
For end_rid_spec :
  • row_hash defaults to the value of row_hash specified in start_rid_spec . If no value for row_hash was specified in start_rid_spec , row_hash in end_rid_spec defaults to 0xFFFF 0xFFFF.
row_uniq The system-generated uniqueness value which is used with the row hash value to uniquely identify a physical row. The uniqueness value consists of two numeric values, usually shown in output screens under column headings u0 and u1.
For start_rid_spec :
  • If row_uniq is omitted, and no end_rid_spec is specified, all the rows in row_hash are selected.
  • If row_uniq is omitted, and an end_rid_spec is specified, row_uniq defaults 0x0000 0x0000.
For end_rid_spec :
  • row_uniq defaults to the value of row_uniq specified in start_rid_spec . If no value for row_uniq was specified in start_rid_spec , row_uniq in end_rid_spec defaults to 0xFFFF 0xFFFF.
= The program will use the most recently saved values for row_partition, row_hash, or row_uniq.

The most recently specified values for row_partition , row_hash , and row_uniq are stored. Separate values are stored for start_rid_spec and end_rid_spec . row_partition , row_hash , or row_uniq . An error is reported if a saved start_rid_spec or end_rid_spec value is used for a partitioned table, but the table referenced in the current command is not partitioned, and vice versa.

For more information on the Teradata File System and physical rows, see The Teradata File System.

For more information on table row structure, see Teradata Vantage™ - Database Design, B035-1094.

Vproc Numbers

In Ferret, the vproc number (vproc_number) is used in the SCOPE command to specify one AMP or a range of AMPs for which the utility performs an action, such as reconfiguration or disk space display.

Valid AMP vprocs have numbers in the range of 0 through 16199.

Map Names

A map includes one or more AMP vprocs. If you specify a map name in the Ferret SCOPE command, the name must be enclosed with double quotation marks. The scope is effectively the same as if you had scoped Ferret to the individual vprocs that are included in the specified map.