2GB Limit on Hash File regarding blink error

Post questions here relative to DataStage Server Edition for such areas as Server job design, DS Basic, Routines, Job Sequences, etc.

Moderators: chulett, rschirm, roy

Post Reply
ggarze
Premium Member
Premium Member
Posts: 78
Joined: Tue Oct 11, 2005 9:37 am

2GB Limit on Hash File regarding blink error

Post by ggarze »

Even if you choose the option "Use directory path", in which the hash fille is now ceated outside the project, does the 2GB limit still apply?
DSguru2B
Charter Member
Charter Member
Posts: 6854
Joined: Wed Feb 09, 2005 3:44 pm
Location: Houston, TX

Post by DSguru2B »

Yes.
Creativity is allowing yourself to make mistakes. Art is knowing which ones to keep.
ravibabu
Participant
Posts: 39
Joined: Tue Feb 13, 2007 12:18 am
Location: vijayawada

Post by ravibabu »

if your data is more than 2 GB ,go to Px jobs ...that's good for your project....
ggarze
Premium Member
Premium Member
Posts: 78
Joined: Tue Oct 11, 2005 9:37 am

Post by ggarze »

Thanks for the responses. Unfortuantely we don't have Parallel extender.
DSguru2B
Charter Member
Charter Member
Posts: 6854
Joined: Wed Feb 09, 2005 3:44 pm
Location: Houston, TX

Post by DSguru2B »

Load your data into a work table (temp table) and do database join. You dont need px just for handling huge reference data.
Creativity is allowing yourself to make mistakes. Art is knowing which ones to keep.
kcbland
Participant
Posts: 5208
Joined: Wed Jan 15, 2003 8:56 am
Location: Lutz, FL
Contact:

Post by kcbland »

The 2 GB limit applies to either the data or the overflow file. You can find that your hashed file uses more than 2 GB if you add the DATA.30 and the OVER.30 together. But, if either reaches 2.2GB, the file will corrupt and the job loading it blow up. You should try to stay under 2.2 GB, but I'm not recommending using an ineffecient file size as a means to stretch the capabilities of 32BIT addressing.
Kenneth Bland

Rank: Sempai
Belt: First degree black
Fight name: Captain Hook
Signature knockout: right upper cut followed by left hook
Signature submission: Crucifix combined with leg triangle
lfong
Premium Member
Premium Member
Posts: 42
Joined: Fri Sep 09, 2005 7:48 am

Post by lfong »

If you do not have Parallel Extender (EE) you can always create a 64bit hash file.
ray.wurlod
Participant
Posts: 54607
Joined: Wed Oct 23, 2002 10:52 pm
Location: Sydney, Australia
Contact:

Post by ray.wurlod »

Hashed files are created by default using 32-bit internal pointers. That is the source of the 2GB limit - it is the largest address that can be accessed with a signed 32-bit integer.

You can create hashed files with 64-bit internal pointers, or convert existing hashed files to 64-bit internal pointers, to overcome this limit. The maximum size of such as hashed file is theoretically around 9 million TB, though most operating systems will not let you create a file this size.
IBM Software Services Group
Any contribution to this forum is my own opinion and does not necessarily reflect any position that IBM may hold.
Cr.Cezon
Participant
Posts: 101
Joined: Mon Mar 05, 2007 4:59 am
Location: Madrid

Post by Cr.Cezon »

You can use 64-bits Hash file. resizing the Hash file created by the job.

execute the TCL command:
RESIZE HashFile * * * 64
before write in the hash file

regards,
Cristina.
ravibabu
Participant
Posts: 39
Joined: Tue Feb 13, 2007 12:18 am
Location: vijayawada

Post by ravibabu »

Hi,

Please split the file useing LP stage .I think,this can be helpful to you.

Please revert back to me if i am wrong.
ray.wurlod
Participant
Posts: 54607
Joined: Wed Oct 23, 2002 10:52 pm
Location: Sydney, Australia
Contact:

Post by ray.wurlod »

You're wrong. Splitting the data stream will not overcome any 2GB storage limit, unless two separate hashed file stages are used, referring to two separate hashed files. Then there's no guarantee that a key being looked up will be in the correct processing stream.
IBM Software Services Group
Any contribution to this forum is my own opinion and does not necessarily reflect any position that IBM may hold.
Post Reply