DSXchange

You could sum across columns in the Extraction SQL thereby avoiding the need for a Transformer stage.

When you're doing the range lookup you need to lie to DataStage about which columns are Key. In your case you need three columns identified as Key: I_USER_MYP, D_DAY_STRT and D_DAY_END. The Columns grid should contain only these and I_EMP_DIM (a total of four columns).

Re-read the second paragraph of my immediately prior post.

If you prefer to work in the project, use DS.TOOLS menu. (This actually installs UNLOCK in the project's VOC file.) Note also that LOGTO is one of the commands explicitly blocked in the Administrator client's Command window. But, once you've used an unlock option from DS.TOOLS, you can then use UNLO...

No size limit for creating File Sets (well, actually there is; not more than 10000 physical files each not more than 2GB on each processing node, but you're nowhere near that!). "Map" suggests that there's some kind of problem with memory capacity, possibly you're sorting or possibly it's in buildin...

1) No.
2) You might be running out of scratch space, temp space or swap space on disk.
3) It creates what it needs. Assuming some kind of heap/merge sort, only some of the heaps need to be on disk at any one time.

You do not need anything in .odbc.ini for connecting to UniVerse or DataStage databases. What you do need is an entry in uvodbc.config specifying that DBMSTYPE = UNIVERSE (and, if a remote database, information about the host and service). Note that the space either side of "=" in uvodbc.config is m...

The input pattern (IP) is indeed unhandled (UP) but there exists a user-defined override (UO) that specifies the bucketing for the pattern ^D+D++T (number, directional, unknown, directional, unknown, unknown, street type). The unhandled data (UD) have been handled by the override.

That's where overrides are useful. If you can figure out that you want to treat all instances of a pattern in a particular way, then you set up a pattern override. But if there are particular exceptions (such as a multi-word organizational name), then you might use a string override for that. The st...

You do not need additional licences to use BCI routines with UNIVERSE data sources. Please post the VOC entries to SDKSequences from both projects. The only obvious reason that the behaviour you describe would occur is that they do not point to the same hashed file. LIST.ITEM VOC 'SDKSequences' in e...

It's perfectly fine to have the user dsadm as a member of any group that has access to DataStage projects.

Typically you make dsadm the owner in those projects (since you're logged in as dsadm when creating new projects), so it's not strictly necessary to have the dsadm ID in the groups.

DataStage server needs to be able to "see" the DB2 client libraries (including DB2 Connect where used). It does not need to "see" the DB2 server libraries.

Well spotted, it's not necessary. I was solving a different problem. The aggregator stage will generate: Groupkey1,Groupkey2,Groupkey3.....Amt1,Amt2,Amt3.... 1111,2222,3333.....$20,$8,$5 2222,3333,4444.....$3,$5,$7 The Transformer stage (adding the columns) will generate: Groupkey1,Groupkey2,Groupke...

I would assume yes. I make it a practice to do so, but can not recall reading anywhere that it is necessary. However, the authentication mechanism within dsrpcd has to pick them up from somewhere, and I believe it only reads these files when it is (re-)started.

Create your own client/server relationship between the two servers (via a "remote UniVerse" DSN in uvodbc.config that specifies host name) and execute the updates only through a routine that uses BCI routines to execute the update remotely.