If SORTDFLT=YES, a SORTDFLT INFORMATION section occurs in the RETIX report before the table and count information.
The line after the SORTDFLT INFORMATION heading contains the following:
The next line, starting with *DSOP CLUSTER, is the count of index entries possible for clustering (DSOP 3). The count is the same as the number of records, a ‘worst case’ estimate unlikely to occur because it presumes that every record has a unique cluster value, which would make clustering a poor choice for this data area.
The next line, starting with *DSOP PARTIAL, is the count of index entries possible for partial space (DSOP other than 0).The count is the number of data blocks divided by 56, a ‘worst case’ estimate unlikely to occur because it presumes that every block group of 56 blocks has at least one block with space for a record.
The next line, starting with *DSOP EMPTY, is the count of index entries possible for empty blocks that have been used (DSOP 3, 4, and 5).The count is the number of data blocks divided by 56, a ‘worst case’ estimate unlikely to occur because it presumes that every group of 56 blocks has at least one block that has been used but currently contains no records.
The next line, starting with *MOVED RECORDS, is the count of index entries possible for tracking moved records. However, because CA Datacom does not save a count of moved records, this value is currently always 0. Because it is normal for this to not be a large number, the estimate is expected to be reasonable for the RETIX estimate.
The last line is TOTAL and provides the sum of the key estimates from the prior lines. It also provides information about what is overriding the SORT= keyword. If the estimated count of keys from the record input is all zeros, the *DSOP values are not added into the total, and the override does not occur.
If the record counts are significantly incorrect but not zero, the sort uses the estimates as they exist. If the RETIX fails because of low sort count issues, change the JCL to provide a more accurate SORT= value then rerun the LOAD without using the SORTDFLT= keyword.
For this example, database 2 was used. The first statement is a RETIX with KEYNAME=*SETR which recounts the records in the area. it was used only to ensure the counts were accurate in the example and is not intended for normal use.
RETIX AREA=DD1,DBID=2,KEYNAME=*SETR COMM OPTION=CLOSE,DBID=2 INIT DBID=2,AREA=IXX RETIX DBID=2,SORT=99999,SORTDFLT=YES, OPTION1='(I)'
The resulting output is as follows:
SORTDFLT INFORMATION TABLE RECORDS KEYS ESTIMATE AGR 308 8 2,464 ALS 1,624 3 4,872 ARA 48 3 144 ATZ 9 2 18 BAS 23 5 115 DVW 37 2 74 ELM 601 5 3,005 FCN 0 3 0 FIL 86 2 172 FLD 11,845 10 118,450 HSD 202 1 202 JOB 0 2 0 KEY 524 12 6,288 KWC 30 9 270 LIB 0 2 0 MEM 0 2 0 MOD 0 2 0 OD 0 3 0 PCV 0 2 0 PER 2 2 4 PGM 0 2 0 PLN 0 3 0 PNL 0 2 0 PRC 0 3 0 PRT 0 2 0 REL 2,118 7 14,826 RPT 0 2 0 STM 0 3 0 STP 0 2 0 SYS 12 2 24 TRG 0 3 0 TXT 93 2 186 *DSOP CLUSTER 0 *DSOP PARTIAL BLOCKS 48 *DSOP EMPTY BLOCKS 0 *MOVED RECORDS 0 TOTAL 151,162 SORT=00000151162
CASO000I SORT FILSZ=E151162,FIELDS=(1,186,BI,A),DYNALLOC=(SYSDA,03) CASO061I RCD IN 00000000,OUT 00149005
In this example, the count of keys sorted is less than the estimate because CA Datacom Datadictionary uses the rarely used NIL key feature to not be part of the index. NIL keys are defined to have a value of either all blanks (x'40') or all binary zeros (x'00').
|
Copyright © 2015 CA Technologies.
All rights reserved.
|
|