FastSort Manual

Sorting From NonStop SQL/MP
FastSort Manual—429834-003
8-14
Configuring a CREATE INDEX Statement
When you specify overflow scratch volumes in a configuration file, the values you
specify override any values in a =_SORT_DEFAULTS DEFINE.
You cannot specify both NOSCRATCHON and SCRATCHON. If you do not specify
either SCRATCHON or NOSCRATCHON, FastSort considers using any volume,
except $SYSTEM and TMF audit trail volumes, for overflow scratch files. Volumes
with less than 1 MB of disk space and volumes protected by the Safeguard product
are also exempt.
SWAP ( swapvol [, swapvol ]...)
is valid only if you specify INDEX or BASETABLE. SWAP specifies the name of the
volume on which to place the extended segment swap file. You can include a node
name in swapvol, as in this example:
$myvol
\nwreg.$sales1
You can specify multiple swap volumes only as DEFAULT swap volumes.
Assigning Default and Explicit Values
You can specify any attribute in a configuration file as either a default or explicit value
for record generator or sort processes. Use the DEFAULT option to specify a default
value. Use the partition option to specify an explicit value.
Use default values when you want FastSort to choose from a set of multiple values.
For example, you might specify scratch volume names or CPU numbers as default
values. You can also use default values to apply a single value, such as number of
records to sort or execution priority, to all processes.
Use explicit values when you want FastSort to use particular values for size limitation
or performance reasons. For example, if one index partition is much larger than others,
you might want to explicitly specify values for that partition.
Default and explicit values are not mutually exclusive. For example, you can explicitly
specify scratch volumes and specify a default pool of CPUs for the sort processes, or
you might specify a default pool of scratch volumes but assign a particular scratch
volume to one partition.
Sample Configuration File
Following is a sample configuration file for loading data from the base table
CUSTOMER into partitions on the index AGEINDEX. It includes both default and
explicit values.
== Sample configuration file for loading index partitions
== in parallel. Creates index AGEINDEX on table CUST, which
== is partitioned as follows:
== $DATA1.SALES.CUST
== $DATA2.SALES.CUST
== $DATA3.SALES.CUST
== \NEWYORK.$DATA1.SALES.CUST
== AGEINDEX is partitioned as follows: