MVSFORUMS.com

[deepa12]

I have coded the below in a pgm
SELECT OCFILE ASSIGN TO OCDD
RECORD KEY IS <ALTERNATE KEY>This is not the primary key

...
Read ocfile , etc is done
as though am accessing the base cluster via primary key

In the JCL
//ocdd basecluster
//ocdd1 <path>

Nomarlly this is not the practice We want to access via alternate key then we code as below:

SELECT OCFILE ASSIGN TO OCDD
RECORD KEY IS <primary key>
Alternate key is <alternate key>
....
read ocfile key is <alternate key>

JCL
//ocdd basecluster
//ocdd1 <path>

If i code like the 2nd option it is taking up a lot of IO
If i code like the 1st option the performance is better & it takes less io

What is the reason.
Actually i want to know how the 1st option works

Thanks in advance
_________________
deepa

[jsharon1248]

Ok. You're still using the true VSAM key in the base cluster when you do your reads with scenario #1. You're only using a different data area in the COBOL program to store the key. Think of it this way, you could change the RECORD KEY data area name to CURIOUS-GEORGE, put it anywhere in the record, and when you populate that field with the value 'XYZ', COBOL will read the VSAM record with the key value = 'XYZ'. What you're doing in #1 is tricky business and I wouldn't recommend it.

Now for the performance issue. Even though you allocated the PATH in your JCL, you never really used it in #1. You didn't specify the ALTERNATE RECORD KEY in your ASSIGN, so you never even opened the AIX. Scenario #2 is using the AIX. You're just experiencing a peformance issue with the AIX. What I'd recommend is increasing the BUFNI. I don't remember the defaults for AIX, they're probably the same as the base cluster. Bump up the BUFNI to 5, 7, or 10, and see if there's any improvement from one to the next.