Hash File Location

[phantom99]

Hi,

It's the beginning of a project here, and we know we will be creating a fair number of hash files that are over 2GB. From my understanding they should be created with 64 BIT addressing as a result, plus I think we'll need to use that hash file calculator to come up with a more efficient hash file.

My Q is:

1) Is it true we can't create the proper hash file indicated above using the GUI? I look slike 64bit isn't even an option.

2)If 1 is true, doesn't the CREATE.FILE statement create the hash file in the project ? We were thinking of creating a seperate directory to hold our hash files in.

3) What are some Best Practices for this, keeping in mind having to change environments down the road from dev to int to prod ? Unless I am wrong, to store hash files in a directory we'd need to create them in the project, then move them ? Must be a nightmare when migrating the project.

Any input would be greatly appreciated, we want to get off on the right foot here.

Thanks ...

[ogmios]

1) Use RESIZE, http://www.dsxchange.com/viewtopic.php?t=86877

2) use mkdbfile: http://www.dsxchange.com/viewtopic.php?t=85457

3) You don't need to create them in the project first. Anyway make a nice initial DataStage job to create and fill your hashfile, and run this job in production to create the hash files for production.

As "best practices/discussion" also have a look at http://www.dsxchange.com/viewtopic.php?t=85364, in the middle you'll find a big piece by Kenneth Bland.

Ogmios

[ray.wurlod]

Even just validating jobs in production will create the hashed files (32-bit only) associated with hashed file stages that have input links. The advantage of this approach is that you don't have to process any data. The disadvantage is that you can't create 64-bit hashed files, so would subsequently need to SETFILE and RESIZE.