DSXchange

what would be the difference between , the oracle bulk loader and normal inserts thru OCI STAGE. sqlldr CONVENTIONAL method is the exact same as INSERT SQL statements, no difference. sqlldr DIRECT method bypasses INDEXES, CONSTRAINTS, and TRIGGERS to write directly to the data files. As part of pos...

As a project administrator, is there any way I can prevent careless users from trashing a project in this way? I keep having to delete files with names like DSLink7 all the time. This happens because the Current Working Directory is the project directory. By default, non-explicitly pathed files end...

Ascential, Ab Initio, Informatica, Sunopsis, MS-DTS, OWB, it doesn't matter, they all use the same methods to load data. The database does all of the work, the tools use the supplied interfaces. Let me be very clear here. Even PX will partition the data to align with the physical database partition...

The database controls what data is allowed to load, not the tool. Please understand this point fully: THERE ARE NO TOOLS THAT EXIST THAT CAN PUT DATA INTO A DATABASE. You either use an ODBC driver, a database supplied interface library (OCI/CLI, etc), or a command line data loader (sqlldr, TMU, isql...

There's a huge difference. You must consider how SQL and relational databases work. DataStage has NO SECRET TRICKS. It simply uses DML statements just like everyone else. Read the SQL statement in the stage and you see INSERT/UPDATE syntax. How you do updates does matter. If your SQL is using wildca...

gpatton wrote:to field in seq file use

(linkname.seq_field_name + 1 - 1)

this will convert it to 0 which will then match with the lookup

I think you mean in the key expression where you do the lookup use:

You cannot do what you want. A hash file is not a database table. The primary key is fixed for a row, it determines where the row is placed in the file and must be known in order to retrieve that row by reference. Using a portion of a multi-column key is nonetheless a different key, and you won't fi...

OCONV(ICONV(link.column, "D-MDY[2,2,4]"), "D-YMD[4,2,2]")

We need to develop a RDBMS independent job to put data into a table, will ODBC stage serve the purose. TIA SAM A sequential text file, probably pipe or comma delimited, is the universal format. You can write your data to a file, it then becomes an exercise in writing the appropriate connectivity jo...

The best tool implemented in the worst way possible will look like a failure. The worst tool implemented in the best way possible will look like a success. 1. The ETL tool will not overcome a poor data model. 2. The ETL tool will not overcome a poor database strategy. 3. The ETL tool will not overco...

Job still failed with ORA-01401 and ORA-00001. First is value to large second is PK violation. If one of your columns is "too large" and its part of the primary key, you have issues outside of DataStage. You cannot load this data into your table because the data does not work. Your problem is not t...

You must have some problem in your korn shell script. I doubt you've killed a job thread. All job processes operate under the program "phantom" and have "DSD.RUN" or "DSD.StageRun" as the next word in the command line. Which of these did you kill?

None more simple and elegant than the way you've done it. THe alternative is to paste your function logic into the derivation of the column, which would probably not fit well given the CASE and/or LOOP statement/ structure you've probably used in the function.

You've probably screwed up your key metadata on the hash file. Check that, because it MUST work that way described if the hash metadata is correct.

If you're using stage variables, you have to have one last row in your input source cause the final group to output. Your job would be easier if you reference the same hash file you are writing. Simply reference the row and concatentate the incoming values onto the referenced row. Then, output the r...