Paralled Extender Compilation Issue

[Veni]

Hi

I have a simple job which contains input as sequential file (1220 fields) , one transformer (using for cleansing the data ) and i am spliting these 1220 fileds in 4 oracle tables.

But while compilation its talking more that 30min .Please suggest me where could be the problem.

Thanks in advance

[ArndW]

Can you compare it to compile times (preferably of jobs without transform stages) on your system at the same time? There is little comparison to be done without a frame of reference.

[pneumalin]

I will try to pivot the file if it is possible, since there is no way you can speed up the compilation time if you have 1220 columns in your PX transform. PX engine literally generates C++ source codes based on transformer content and pass it to associated C++ compiler for compilation, and that source code must be big. You can look up the source code in associated folders ofthe project.

[ArndW]

Actually the compilation time can be greatly reduced by using RCP if that fits the design of that job; but even a small job with a transform or two can take ages on an overloaded system; I've had jobs take over 10 minutes to compile in the past where the computer didn't seem overly busy.

[Veni]

Thanks for your comments , I will try with that option

[Veni]

Do we have any limitation on writing expressions in Parallel Extender Transformation Stage ,because when i am using 300 colums with same expressions my job is working fine and compliation is also fast but when i am using all 1220 colums in single transformer , its throwing lot of errors (Error when checking composite operator :Function Decimal_from_String)

Please help me in understanding the issue.

[ArndW]

Take your job that is not compiling and start removing columns until it does compile. Either there is a specific column that causing the compile to fail or it is the amount. You will need to narrow down the exact cause more before useful help can be given. If the error is because the transform stage has too much code in the derivations then it is a matter for IBM/Ascential support; otherwise it has something to do with the last derivation you removed before you got a version that would compile.