Truncated flat file

[Former Member]

Hello Experts,

I have a flat file with almost 244 columns. When I schedule it for upload - the preview looks fine but the PSA shows just 87 columns. The data target shows blanks in the records which are not displayed in the PSA. What do you think is the problem?

Any kind of ideas will be accepted.

Thanks,

Anjali

[Former Member]

Note: 130253.

[Former Member]

Hello Guys,

I got the following statement from the SAP note:

"when loading large data quantities from a file, we recommend that you split the file into several parts. We recommend using as many files of the same size as there are CPUs. These files can then be loaded simultaneously into the BW system in several requests. You need a fast RAID for this.".

So does it mean that I need to truncate my flat file for ex: I have 244 columns into 122 each, name them as 1 and 2 and upload them using the option PSA and data targets in parallel and then upload the data.

Any suggestion would be greatly appreciated.

Thanks,

Anjali

[Former Member]

I think the note essentially means that split the number of records. If you have 10 Million records and 5 CPUs on ur system. Then split them into 5 files each of 2 million records.

Hope that helps.

[Former Member]

Thanks Sandhya,

I realised that late :). But what do you think is creating my problem. I mean truncating my flat file which has about 244 fields but just few records probably a couple hundred.

Let me know if someone else faced a similar problem.

Anjali

[Former Member]

Hello Guys,

Probably my question was vague. I have a flat file which has 244 fields and just about 180 records or so. Whenever I try to look at the preview or upload the data, just half of the fields are shown. For example out of 244 fields I can just see 122 fields i.e. part of my flat file is gone. Is there any setting to upload all the fields. There is a limit of 249 but my flat file has just 244 and also the flat file has about 650-700 bytes and doesn't exceed 1000. So guys please anyone who knows a clue of how to go about this, will be greatly appreciated.

Thanks,

Anjali

[somnathkumar]

Hi Anjali,

Firstly, the number of records in your file is not a problem at all. Your limitation is the length of one record. See question 3 in note 545923. It says that your transfer structure cannot be more that 1962 bytes including the system fields.

Now to calculate how many fields will fit into 1962 bytes. For this, take the first field. If the infoobject is defined as CHAR 9 then it is 9 bytes, a key-figure is 10 bytes most of the time. So add up. Remember even if in your flat file there are 2 chars but it is mapped into a char 9 field, it still counts as char 9. I don't think number of columns is a factor.

You can split your file into multiple files of 1962 bytes size but keeping the unique key of the records common across all the files. So that you can merge then in an ODS maybe. Again remember that fields like request number, date packet number, partition ID and record number has to be there in every PSA and a part of the 1962 limit.

Tell me if this helped you as I currently do not have access to a system.

[Former Member]

Hi Anjali,

If you are sure that number of fields and a total record length is not a reason, then I'd give you a recommendation to make sure that your communication structure has all 244 fields and the appropriate transfer rules.

Best regards,

Eugene

[Former Member]

Hello Eugene,

I am not sure but if you can clarify my doubt probably I can go ahead with your recommendation. I have 4 infosources from the same source going into a common ODS and then into related cubes. Ex:zqm_xx_pn,zqm_xx_sip,zqm_xx_hf,zqm_xx_ami (infosources) going into a common ODS zqm_o01. I have just 120 fields in 'pn',244 fields in 'sip', 200 fields in 'ami' and about 104 fields in 'hf'. I have a problem with even 'hf'. I can see just 96 or 99 fields and i miss the last 4 or 5 fields in the smallest infosource file. So please if you can understand my question see if the recommendation works or is their any other solution.

Thanks again,

Anjali

[Former Member]

Hi Anjali,

Let's make sure that your record length doesn't exceed 1962 bytes for the smallest IS - 'hf'.

1. Define the name of the PSA table for this IS. - RSA1 - PSA area at the left - find your IS - right click - Delete PSA Data - scroll the upper frame to the right - find the PSA table name (DDIC table of the PSA column).

2. Tcode SE11 - put the PSA name found - sum up all the fields lengths.

Let's know what you'll get.

Best regards,

Eugene

[former_member188975]

Hi Anjali,

So does the preview show you all 244 columns of data? Can you check the structure of the PSA table in SE11..does it have all the colunms?

SAP mentions: "A table may not have more than 249 fields. The sum of all field lengths is limited to 1962 " ...Could this be the problem with your data?

[former_member183113]

Hai Bhanu

what is sum of field length 1962??? can u elaborate this??

Regards,

Ravi kiran.