DSXchange

You are correct. These are concepts within the Universe database. The DICT side of a hashed file or "table" stores the field names or formulas. The dictionary items are evaluated when you execute SQL or run the old fashioned reports called Recall or English within Universe. The VOC is the system tab...

I am sure Ray is correct. IBM owns it and will not give it out.

A true batch job is what was used before sequences. That is what VMark originally called them. A batch job is a job which has no stages. It is all job control code.

You are correct if this is not what they meant then my answer is wrong.

Search on DEFFUN. This has been covered a lot. There are lots of examples in EtlStats and on this site.

Abort was a poor choice of words. It should skip to the end of that sequence. So each sequence should check the audit table to see if it has been run today. If it has then it exits to the last step. If it has not been run today then run all of the jobs normally. So do this check first in the sequenc...

The surrogate key concept is cleaner. Then all the keys that made up the uniqueness of the original fact record and are represented in the surrogate key. So whatever is less work or works better either you keep adding dimension keys from the original fact table or you add a surrogate key to it. You ...

Q1 First one Q2 You need a job or a routine to check for a finished date. If there is a finished date then the job or routine needs some way to signal the sequence that this step is finished. I would put this in as the first step of the extract sequence. I would either abort the job or return a @FAL...

Sure, you could have 2 fact tables. One at the current level. The other just has the statuses plus the minute id. You need to join your 2 fact tables at the day level. So you would also need the day_id. You need to figure out what other keys you need to make join work. You will also need an index on...

dsjob -report <project name> <job name> XML

Will give you link names and row counts and lots of other fields.

Here is the part you missed: If they show acef 010c (or 010c acef on an Intel CPU), then it is a 32-bit hashed file. If they show acef 020c (or 020c acef on an Intel CPU), then it is a 64-bit hashed file. For static hashed files, omit the "/DATA.30" from the hashed file pathname. So maybe both hashe...

If you are trying to rethink the idea of a cursor then think of the primary input of a job as the cursor. Next you can do lookups of other things like a stored procedure would. I think of jobs like they are visual representations of stored procedures. Maybe that will help you transition to the DataS...

Nick I think Ray is assuming that because you marked your profile as parallel then you want to build a PX or parallel job and not a server job. Both are part of DataStage. PX which is also called parallel is the newer part of DataStage development. It can be many times faster than a server job. It t...

The cache size is controlled in the DataStage Adminitrator. It is a property of a project. If you exceed the size then you need to reduce the amount of data you put in a hashed file or change this property. You can reduce the amount of data by only having columns you need. So eliminate unnecessary c...

I have seen it take days or weeks to do a historical rebuild.

That email was for Ascential. If he still works the email will be an IBM one.