DSXchange

Imagine that the SELECT statement that defines the view selects 10 million rows, and when you select from the view you want only 10,000 of these. Selecting from the view means that the first, large, SELECT has to occur to materialize the view, and only then can your second, small, SELECT proceed. If...

Yes. But the cost of materializing the view may be one that you wish to avoid.

Have you written a parallel version of the Fmt() function? If so, please post it.

Meanwhile, you could use something like converting to a string and overwriting the rightmost part of "000". But, in parallel jobs, integers do not have leading zeroes - only strings can.

If you can pre-sort the data by the grouping columns, and change the Aggregator method from Hash to Sort, you should not encounter any memory issues.

It is not.

Start with a plan. Business analysts might refer to this as a "source to target mapping" or a "target from source mapping" document. The latter is easier for an ETL developer to use. That document becomes your specification. You can then plan the stages and functions you will use to implement a set ...

You didn't need the Convert() function - you could have specified the "-" as the delimiter in the Oconv() function.

To complete the answer, including the bits the ever-hasty 2B left out: Login to UNIX change directory to $DSHOME (which I assume to be set) execute the dsenv script execute the bin/dssh command to enter the TCL environment (the prompt is > ) LOGTO projectname DS.TOOLS invokes a menu for administerin...

Welcome aboard. :D

Verify that your umask is 002, not 022, so that anything you create will be RW access for the remainder of the members of your group. Verify also that the permissions on the objects in the project is the same as that of the project itself.

Welcome aboard. :D Etiquette Note We do not do "urgent" or "ASAP" here. This is an all-volunteer user community - nobody gets paid for posting anything. If you need urgent answers, sign up with your support provider for premium services, and learn the true cost of "urgent". Search the forum for DEFF...

The total amount of time will probably be about the same - after all, the same amount of work has to be done. If the Informix table has an index on the grouping column, then grouping in the database may be faster than grouping in an Aggregator stage. Conversely, the Aggregator stage forms its groups...

"UTF8" is like "UNIX" - every vendor has their own idea about what it should be. There are many eight-bit encodings of Unicode, all of which are entitled to call themselves "UTF8". There is no single UTF8.

Now check the following four environment variables in your DataStage Administrator, to ensure that they are correctly set.
APT_COMPILER
APT_COMPILE_OPTIONS
APT_LINKER
APT_LINKER_OPTIONS

Aaarrggh! Use meaningful names. Don't use the default of using the link name as the hashed file name! I wish they'd never introduced this default. It can lead to too many problems, not least wasted disk space.

Don't believe everything you read about "performance", which is ill-defined in an ETL context. There are three stages that effect horizontal combination of data, each subtly different - Lookup, Join and Merge. Only the Lookup stage allows conditional lookups. The Lookup stage has a single reject lin...