why DFSHDC40 Randomizer module is so much slow?

Pepero2k3 · New User Joined: 07 Sep 2008 Posts: 4 Location: Iran

On an online environment I've a HDAM DBD in which 80% of all my operations on it is the insertion.
In the 20% remainder i read it qualify(qualify GU).
Also the Key is a character string with the total length of 33 bytes.
The total number of the database records would be 5,000,000 records in a day.
I've considered the Os access method as OSAM and the database block size of 4k. The HDAM related parameters in the DBD definition is as follows :
DBD NAME=RETDTLRI,ACCESS=(HDAM,OSAM),
RMNAME=(DFSHDC40,20,100000,),PASSWD=NO
DATASET DD1=RETDTLRI,DEVICE=3390,SIZE=4096,SCAN=3
SEGM NAME=TLRINFO,BYTES=60
FIELD NAME=(TLRKEY,SEQ,U),BYTES=33,START=1,TYPE=C
DBDGEN
FINISH
END
The problem is that the insertion operation is so much slow.
what shall i do?

Bitneuker · Posted: Tue Sep 09, 2008 12:19 am Post subject:

Are the keys to be inserted in sorted order?

Pepero2k3 · New User Joined: 07 Sep 2008 Posts: 4 Location: Iran

Hi
The insertions take place in a high rated online transaction based environment. But the interesting subject is that in an empty databse with a single insertion transaction the response time is undesirable.
I've an experience with the following parameters
RMNAME=(DFSHDC40,5,400000,),PASSWD=NO
DATASET DD1=RETDTLRI,DEVICE=3390,SIZE=1024,SCAN=3
but the problem was still alive.
But when i use the DFSHDC10 instead the insertion response time would be desirable. But due to the key length i cannot use the DFSHDC10.