Write and Read the same hash file in same job?

[Desigal24]

i am writing to a hash file first and in the second transformer i read from it for a lookup. I dont have any caching enabled for the hashfile.
Will this work? will all the records be written successfully before the read operation begins?
ANY HELP IS APPRECIATED.

---------- ------- ------- -----------
ORAOCI ---> TXFR ----> TXFR ----> Seqfile
---------- ------- ------- -----------
| /\
V |
----------- -----------
Hash file Hash file
------------ -----------
Thanks,
G

[chulett]

We really should have a better picture of your job to give you a proper answer. Wrap your ascii art in 'code' tags so that spacing is preserved. Make liberal use of the 'Preview' function to ensure it all lines up properly. Keep at it until you get something usable.

Then we can talk.

[ray.wurlod]

It will work, provided you don't use caching for read or write. Also enable lock for update. It is essential that you use separate Hashed File stages; add an annotation warning future developers that the two stages refer to the same hashed file (not hash file).

[chulett]

Code: Select all

-------    -------     -------     -----------
ORAOCI ---> TXFR  ----> TXFR  ----> Seqfile
-------    -------     -------     -----------
            |              /\
            V              |
         -----------  -----------
         Hash file    Hash file
         -----------  -----------   

Ok, there's my hack of your ascii art.

Depending on what you mean by 'work' the answer is either yes or no. Yes, a record written in the first xform will be available for lookup in the second - if you follow Ray's advice.

If work is related to your second question - then no. DataStage will process a record from source to target before the second record starts its journey. So, no, not all records will be written to the hashed file before the read operation begins. Is that what you need? All writes before all reads? Not sure that makes sense, if that's what you need please explain your requirements...

[ray.wurlod]

Usually there's an extra step, to pre-populate the hashed file from the target table.

[sb_akarmarkar]

i am writing to a hash file first and in the second transformer i read from it for a lookup. I dont have any caching enabled for the hashfile.
Will this work? will all the records be written successfully before the read operation begins?
ANY HELP IS APPRECIATED.

---------- ------- ------- -----------
ORAOCI ---> TXFR ----> TXFR ----> Seqfile
---------- ------- ------- -----------
| /\
V |
----------- -----------
Hash file Hash file
------------ -----------
Thanks,
G

Yes, It works better dont enable caching.

Thanks,
Anuapm

[chucksmith]

Do you think the order of the output links in the first transform stage is important? If so, I suggest you order your output links to ensure that the write to the hashed file occurs before the passing of the row from the first transform to the second.