DSXchange

Fairly certain. The point about pre-loading the hashed file to memory is that it pre-loads the page (group) structures, so that the hashing algorithm does not change - it still computes the address of the group in the hashed file, only it's now a memory address rather than a disk address. This is th...

That behaviour will be governed by whether you've enabled automatic error handling. You may need to disable this, and handle (or not handle) failed job activities explicitly in your job sequence.

Parallel routines do not undergo the same "catalog" mechanism that server job routines do, hence DEFFUN declarations are inappropriate. Parallel routines are not readily called from job control code, since parallel routines are written in a different language (C++) from job control code (DataStage B...

You can have as many -param name=value options as needed.

You can't control the columns (fields) that are loaded into memory. All you can do is load that subset of columns into a different hashed file, and cache that hashed file. Thus there would be no problem of the kind you fear in point 2.

The crux of the error message is "Unable to create operating system file". You have to figure out why this is - permissions, non-existent directory in the pathname, whatever.

If you're passing a parameter to the job you have to pass it from somewhere - perhaps a job sequence, perhaps a shell script. It's there that you generate yesterday's date. In a job sequence a suitable expression is Oconv(@DATE - 1, "D-YMD[4,2,2]") (Adjust the specification to suit...

I don't believe that the cache space for OSH being insufficient would lead to the "broken pipe" error thar was reported. This looks much more like a timeout possibly caused by too many processes on the machine (and therefore too long a wait to start another process).

If you do not specify the partition table, then your SELECT will operate in sequential mode. If there is an Oracle table that specifies that your CUSTOMER data can be read four-way parallel then naming that table will allow DataStage to take advantage of it.

HFC will generate a command to create a 64-bit hashed file (note: "hashed file" not "hash file") if you select the appropriate option.

DataStage server job expressions do not use SQL wildcards. DataStage server job expressions have a more flexible set of pattern matching capabilities which you can read about in on line Help (topic Pattern Matching Operators) or in the DataStage BASIC manual.

Hi ray, Thanks for your reply, can I resize dynamic 30 file to 64 bit and I read in one of the post that there is HFC.exe in client version of datastage. How can I use the HFC to see the type, modulus and seperation (I dont have HFC.exe with me) is there any other way than HFC and I have three key ...

A couple of corrections. HFC does not actually create the hashed file; it generates a command that allows you to create the hashed file either from DataStage environment (a CREATE.FILE command) or from the operating system environment (a mkdbfile command). It does not calculate the average record si...