Dumping Hash File

dickfong · Post by **dickfong** » Thu Feb 06, 2003 10:22 am

Is it possible to dump the content of a hash file to a sequential file using scripts (DS Basic or UNIX shell script)? What I want to do is sth like a database export of table. Many thanks

chulett · Post by **chulett** » Thu Feb 06, 2003 3:00 pm

Sure. I can't supply something already written, but it can certainly be done in DS Basic. Very high level:

Use OPEN and OPENSEQ for the hash and flat file, respectively. Loop thru the hash with READNEXT and write to the flat file using WRITESEQ. Also be aware of WEOFSEQ and CLOSESEQ. If you don't have hard-copy manuals, there should be pdf versions of all manuals (including the BASIC manual) that installed with the client software. Also check the online help. [:)]

-craig

ray.wurlod · Post by **ray.wurlod** » Fri Feb 07, 2003 3:05 am

Where and why do you want to do this? In particular, do you want the data or the metadata from the hashed file to be dumped?

Ray Wurlod
Education and Consulting Services
ABN 57 092 448 518

dickfong · Post by **dickfong** » Mon Feb 10, 2003 8:12 am

Thank you both of you.

What we are trying to do is to create some backup points of the data staging area in between our programs so that we can re-reun the programs from those backup points in case of faliure.

The idea is like we have hundreds of jobs all linking by 2 job sequencers, the 2 sequencers would be executed in sequence and in between the 2 sequencers we would like to insert a job to backup all the intermediate hashes and sequential files so that in case of faliure of part 2 sequencer, we can restore the backup files and restart part 2 instead of the whole ETL process.

Any idea of how it can be achived?

Many thanks

ray.wurlod · Post by **ray.wurlod** » Mon Feb 10, 2003 3:22 pm

If you're going to insert a job to do this, why not list DataStage handle it? Source is hashed file, target is sequential file stage (or FTP stage if you want the backup on a different system).

Ray Wurlod
Education and Consulting Services
ABN 57 092 448 518

dickfong · Post by **dickfong** » Tue Feb 11, 2003 9:49 am

Our concern is on the performance. Backing up with scripts (UNIX or UV) may generate less overhead than by a DS job, can I say so?

Another concern is the effort to develop hundreds of jobs which perform the same task. i.e. to copy from hash to sequential file or vice verse, with different table definition.

Please advices. Thanks a lot.

mnevotti · Post by **mnevotti** » Tue Feb 11, 2003 2:26 pm

Couldn't you create a shared container in which you have the source as a hashed file, target is a sequential file using FTP? That way the other jobs can just use the shared container and the same task doesn't need to be developed hundreds of times.

-Melinda

ray.wurlod · Post by **ray.wurlod** » Tue Feb 11, 2003 5:51 pm

How will you identify the hashed files?
If they're all in a folder (ideally with nothing else) you could use a folder stage to feed the file names into a job that would process them all.

Ray Wurlod
Education and Consulting Services
ABN 57 092 448 518