DSXchange

After weeks of investigation we found that server jobs were way faster because they were using OCI stage. This issue comes out of the fact that it is not possible to specify the array size for update as it was for OCI with server. In Server, when i try to put an array size of 1 for update, the perfo...

You would send them the query, the explain plan and the table ddl and the Server query and explain plan as well I would imagine. Let them take it from there. No, there's no configuration step or Oracle options you're missing here. About to open the case... I've made some little progress in the mean...

I receive the following error whenever I try to view data in an iput stage: Error calling subroutine: DSD.Browse (Action=5); check DataStage is set up correctly in project pricing_dev Warning: Data that exceeds the maximum display length has been truncated Could you explain a bit more. Does this ha...

OK. I don't know how they would be able to reproduce it. Guess there is no other option now. I was just meaning that maybe there is an admin configuration to be done on the OE stage that was not done correctly. Also, do I need to put some specific Oracle options on the table (PARALLEL, etc) ? Just t...

Sorry mister

I meant the OCI Oracle Stage in Server against the Oracle Enterprise for Parallel

Hi Craig,

My official support is ... IBM. I don't know if they can help me in this case but I can try.

Do you think the difference could be cause by a misuse of the OCI stage (in EE we use Oracle Enterprise).

I wonder if there is a configuration file of some sort.

Wow, I tried the same job with a server method and it ran so much faster.

I've read this issue here befor but it seems so crazy to me...

How can Server be faster than parallel ??

Will try...

Does this problem rings a bell to someone ?

Hi,

I've just found something interesting:

My update key is called INV_BK, hence I'm passing ORCHESTRATE.INV_BK in my where clause

If i edit the update statement by replacing this value by a constant, the explain plan becomes correct and it used the index !!

Hi Ray,

No it's very fast.

The problem comes out of the Database for sure

it performs a full scan with a cost of 40,000 whereas it should do an index row scan (the cost would be only 4 !!!!)

Dataset are internal files which support parallel storing. Not human readable. Sequential files are text files that are human readable but can only be treated sequentially.

Hello people, I have a simple job which retrieves 5000 update records out of a dataset and apply them to a target Oracle database using an update only mode. It is extremely slow I'm running the update statement on a single key which has a unique index on it. It has been also previously analyzed. I t...

When combining a range lookup with a regular lookup, you always have to trigger the multiple rows option whatsoever.

Hmm Don't know much how to help but here are my two cents... The range function should be done this way (checking one code + 1 date against list of codes and values interval such as dates for instance) 1) On the code : Set "=" and join with the code On the start/end values, set "a..z&...

Couldn't have answered better