DSXchange

greggknight wrote:The transformer is the easy way out but the most costly in terms of performance.

I don't believe that's true any more. But I have no empirical results either way.

That error suggests that there's a value in your data for which you have not coded. Try adding an "otherwise" case. Are you directing particular values to particular output channels?

32-bit or 64-bit Windows?

At least one node must be in the default node pool (""). Whenever you create a configuration file, use the Check facility to make sure it's one that will work. This can be done in the Configurations manager from the GUI or using the orchadmin -check command.

Things changes in version 8. Roles are stored in the common metadata repository, not in the Template.

chulett wrote:In a Server job, one can define their own collating sequence in a Sort stage so that it knew to sort H then D then T. Is there an equivalent on the PX side?

No.

Those utilities probably won't solve this one. Your first task is to determine why the Administrator is reporting that the project still exists, and to tell us who deleted it incompletely and using which method. If there is an entry for the project in the UV.ACCOUNT table, this entry can be deleted....

Sort by original order (or preserve the sorting throughout). Maybe also sort by Convert("HDT", "123", InLink.RecordType).

Text files don't really have data types at all (they just have text). Therefore it is not possible for them to contain null. The Sequential File stage allows you to specify how "null" is represented in data, both globally and at the individual field level. This does not mean that null now ...

So your code destroys the control records.

It's an easy enough change to read them before CLEAR.FILE and reinstate them afterwards.

ISO8859-1 will not handle Chinese characters. You need one of the maps that will.

Use the Performance Analysis tool in DataStage Designer.

The configuration does not contain the invoking node dstindiai36.

You can not invoke a parallel job using a configuration that does not include at least one node on the machine from which the job is invoked. (Obviously grids are different, but you made no mention of executing on a grid.)

I'm with Craig. 15GiB of data in one file is still going to occupy 15GiB in multiple files.

How come everything you post is "weird"? There's not enough information in your post to help you. Give examples of the query, and what you consider to be junk rows, and why. You would expect weird results using a parallel ODBC stage in a server job, so why did you post a question about a p...