Search found 53125 matches

by ray.wurlod
Tue Sep 25, 2007 4:41 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: ABOUT HASH PERTITIONING
Replies: 5
Views: 1707

Hashing is a widely-used technique for reliably (repeatably) selecting one from a finite number of alternatives based on a given value. In parallel jobs, hash partitioning chooses one from the finite number of processing nodes based on the combination of values provided as "key" on the partitioning ...
by ray.wurlod
Tue Sep 25, 2007 4:35 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Schema definition
Replies: 8
Views: 3542

:idea:
Whenever you see "internal error" you need to involve your support provider and, through them, the vendor.
by ray.wurlod
Tue Sep 25, 2007 4:31 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Insert then Update & Update then Inser
Replies: 4
Views: 11869

I disagree about your assertion about "too large" - any lookup will be aided by the primary key index, or you can pre-build something like a Lookup File Set. As to your actual question, you just have to wear the slight difference in efficiency on those days where there are more updates than inserts....
by ray.wurlod
Tue Sep 25, 2007 4:28 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Error - Importing DATE metadata from Oracle in DS801
Replies: 1
Views: 1926

That's because it's a timestamp. There is no automatic conversion to date - that's what the error message is telling you. Therefore you can not simply assert that the column is of type DATE and expect to import from an Oracle TIMESTAMP into it. Put another way, your metadata must accord with what's ...
by ray.wurlod
Tue Sep 25, 2007 4:25 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Alternate Lookup
Replies: 1
Views: 795

Why?

Your specification is quite specific about needing one lookup and a second conditional lookup. Therefore you need two. You can, of course, make the second one conditional.
by ray.wurlod
Tue Sep 25, 2007 4:24 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Repartition Dataset
Replies: 1
Views: 898

You can not "repartition the entire Data Set" without deleting and re-populating it. Therefore you have to ensure that the appended rows are partitioned identically to the rows that are already in the Data Set.

This is the same answer as the one I gave yesterday.
by ray.wurlod
Tue Sep 25, 2007 4:21 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Import Table Definition (ODBC) from Datastage Manager Error
Replies: 10
Views: 5035

Have you checked and double checked that you are providing the correct user ID and password? After all, the error message referred to an authorization specification, rather than to a connection issue. Are you doing anything fancy with the ODBC connection string?
by ray.wurlod
Tue Sep 25, 2007 4:19 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Conversion error calling conversion routine timestamp_from_s
Replies: 10
Views: 6118

Sequential files don't have/enforce data types - they only have "text". In the ELSE side of your IF expression you are providing no fractional seconds whereas in the THEN side you are providing five. No wonder Oracle is confused. What is the data type of this column in the Oracle table, as imported ...
by ray.wurlod
Tue Sep 25, 2007 4:15 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Warning in Aggregator stage
Replies: 3
Views: 1499

U (a different poster) may not, but will siva7143 ?

Correct spelling of the second person personal pronoun is "you", not "u".
by ray.wurlod
Tue Sep 25, 2007 4:12 pm
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Warning while loading data
Replies: 0
Views: 1041

Welcome aboard. It's time to become a detective. NOCHECK appears in the control file for sqlldr that is generated by the Oracle Enterprise stage - you can find and view this file in the directory you specified in the stage. Now backtrack into your design and find the stage property that causes this ...
by ray.wurlod
Tue Sep 25, 2007 6:01 am
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: How to do left outer join and right outer join in DS 7.5.2
Replies: 4
Views: 3511

Hi.. Thank you very much for your solution. But am confused, how can we do left outer join using transformer stage?? Do we need to check for Isnull condition in constraint division?? The natural behaviour of the Transformer stage is to perform a left outer join; NULL is returned from the "right" (r...
by ray.wurlod
Tue Sep 25, 2007 5:55 am
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: Modify Stage Question - Multiple conversions
Replies: 7
Views: 3065

It is documented that two adjacent Modify stages is not supported. Place a Copy stage between them if you need two.
by ray.wurlod
Tue Sep 25, 2007 12:44 am
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: problem with string to decimal conversion
Replies: 4
Views: 5965

Yeah? What about 130 and 1460?
by ray.wurlod
Tue Sep 25, 2007 12:43 am
Forum: IBM<sup>®</sup> DataStage Enterprise Edition (Formerly Parallel Extender/PX)
Topic: How to do left outer join and right outer join in DS 7.5.2
Replies: 4
Views: 3511

Welcome aboard. Assuming this is a parallel job, since you posted in the parallel forum, you use a Join stage. If it is a server job you can only do a left outer join, using a Transformer that has a reference input link. However, it is easily provable that reversing the sense of left and right allow...
by ray.wurlod
Tue Sep 25, 2007 12:37 am
Forum: IBM<sup>®</sup> Infosphere DataStage Server Edition
Topic: Routines and Subroutines
Replies: 2
Views: 831

Routines in DataStage server job is a class that includes before/after subroutines and transform functions.

Put another way, Routines is the name of the root node in the Repository in which before/after subroutines and transform functions are stored.

At least prior to version 8.