Coding the ESTIMATE command and keywords

TYPE

PHDAM, PHIDAM, PSINDEX, HDAM

Input database type

Specify the type of estimation that you want to perform.

COUNT

1-1001

None

Specify the number of partitions to create.

If the COUNT keyword is specified, partition boundaries are determined after all input records are read. The total number of database records is divided by the COUNT value to determine the number of records in each partition, and partition boundaries are established within the input records.

Mutually exclusive with the RECORDS, PARTSEL, and KEYSIZE keywords.

SIZE

1-4000

(The former value is approximately 20 cylinders; the latter value is approximately 4 GB.)

If you specify OSAM8GB=Y, the valid values are 1-8000 (see the OSAM8GB keyword below).

2000

(approximately 2 GB)

Specify the size of a partition in multiples of 1 MB.

The SIZE and COUNT keywords are mutually exclusive. If neither keyword is used, the SIZE default is SIZE(2000). If both keywords are specified, the SIZE keyword is used and the COUNT keyword is ignored.

If the SIZE keyword is specified, records accumulate until the threshold is reached. The root key is saved at each threshold, and values for HDAM RAPs and blocks are calculated. Segment sizes are adjusted to accommodate new extended pointers. The free space requirement that is specified in the DBD is honored.

For HDAM databases, the number of blocks in the RAA is increased to honor the HDAMGROW or the DBGROW value. A unique set of HDAM RAPs and blocks is calculated for each partition.

Mutually exclusive with the RECORDS and PARTSEL keywords.

RECORDS

1-2147483647

None

Specify the number of database records to place in each partition.

Mutually exclusive with the SIZE, COUNT, PARTSEL, and KEYSIZE keywords.

KEYSIZE

1 - [keylength minus 1]

Keylength

Split a single partition into 'count' partitions, which are based on a partial root key. This action might be useful if the key contains specific entities (date, state, ALL type, and so on) and you want to split the database into independent partitions on this key component.

MERGEDSG

Y

N

N

Specify whether to merge any multiple data set groups to create a HALDB that has a single data set group (Y); all input data is accumulated as if no secondary data set groups ever existed; or to preserve the data set group format (N); only the data from the primary data set group is accumulated, and all other data set group data is ignored.

HDAMGROW

10-50

20

LASTFREE

Y

N

N

Specify whether or not to leave the last partition empty for future growth (Y) or fill the last partition with records in the same way as the other partitions (N).

For example, if COUNT(10) and LASTFREE(Y) is specified, the number of input records are spread across nine equal partitions leaving the tenth partition free. The LASTFREE keyword is valid only when the COUNT or the RECORDS keyword is specified. If the input database is a partitioned database, the last partition in the range of partitions to be repartitioned or the partition to be split must be the last partition in the input database so the high key of the empty partition is high-values. Otherwise LASTFREE is ignored.

PARTMASK

(namePattern, sequenceType, sequenceStart, sequenceIncrement)

namePattern = first six characters of DBD name

sequenceType = alphanumeric

sequenceStart = first characters of sequence type

sequenceIncrement = 1

Specify the partition name prefix.

The PARTMASK keyword is used by DBRC to create the partition ddnames. You can specify a maximum of seven alphanumeric characters. IMS uses the eighth for the data set group.

For more information, see Coding-the-PARTMASK-keyword.

DSNPREFX

1-32 alphanumeric characters

MODIFY.MYNAME

Specify the data set name.

IMS appends the HALDB partition suffix to the specified data set name.

NOPINIT

Y

N

N

Specify whether a CHANGE.DB PART(partName) NOPINIT command follows each INIT.PART or CHANGE.PART specification in the DBRCOUT data set.

PARTSEL

PSEModuleName

None

Specify the name of a PSE module to test and determine how it performs with the data from the database.

The module must be available in the STEPLIB concatenation of the ESTIMATE step.

You must provide a data set with ddname BMCRECON containing the DBRC control statements to define all the database partitions.

The utility uses the partition names, DSNPREFX value, and KEYSTRNG values from the BMCRECON data set. It passes the key of each database record to the PSE to determine the partition in which it belongs and reports the sizes and counts for each partition.

REUSPRTS

Y

N

N

Specify whether to create a DBRC CHANGE.PART statement instead of DELETE.PART and INIT.PART statements when the PARTMASK option generates a partition name that matches a partition name in the input database. Also specifies whether to use the existing partition number to create the data set names in the IDCAMS delete and define statements.

OSAM8GB

Y

N

N

Specify whether 8 GB OSAM partition data sets can be modeled.

This keyword is valid for Version 14.1 and later of the IBM IMS system, and for Version 13.1 with APAR PI23918 installed.

This keyword requires Recon MINVERS of 13.1 or higher. If the MINVERS is 13.1, the Cross-DBRC Service Level ID (CDSLID) must be 2.

Advisor includes commented CHANGE.RECON statements for MINVERS and CDSLID in the DBRC output data set, when appropriate.

When you specify Y, the accepted values for the SIZE keyword are 1-8000.