IBM Mainframe Forum

Hi All,

I have a Transaction Data file with 5 million records. And the key information of the file is as follows.
It holds the data for a calendar month and the data gets appended every day mid-night

LRECL = 9855
RECFM=VB
DATE field starts at position 3. And it looks like 'YYYYMMDD'.
For Example, Transaction Data file for November 2015 has the records starting from 1/Nov/2015 till today. latest data is at the bottom of the file.

I have 2 scenarios here.
One is to extract the data from the input-file (5 million records) for a specific date, mostly yesterday (1 day before current day)

I use the following code


Second is to extract the data from the input-file (5 million records) starting from a specific date.

I use the following piece of code


Query: Above piece of codes are working fine to my requirement. But these are processing 4.8 million records just to extract 20,000 records unnecessarily. Is there a better way of doing it.

You can use GT or GE, LE or LT, express ranges or groups with AND and OR.

However, what magic do you feel should work to get your task done without reading the records prior to the ones you want?

Hello,

Analyst_Kumar wrote:I have a Transaction Data file with 5 million records.
....
It holds the data for a calendar month and the data gets appended every day mid-night
....
....

Is there a better way of doing it.

In continuation to what Billy has already mentioned, change the process-design a tad bit.
Fwiw, it is a li'l weird that we first append data right at the end of a file, and then the very next day go ahead and extract the very same data, again - from the end of the very same file.
A possible solution would be to simply write a GDG version; from your existing process; and later append this GDG version to your transaction file. Have GDG limit set to 2; this way you are not eating up much additional space, also you have two day old data readily available; this removes the need to run one of your SORTs altogether.

For the second case, wherein you have a few random dates, like Billy said - you dont have much to go on with, other than reading the file and figuring out if the date is the one you need.

Hth.