How splitting happens in PI

[Former Member]

Hi friends,

how file splitting happens in PI.

Is there any different ways to do splitting the file.

eg-if send the file with 50mb how the file will splite in pi

if it is record type how or if it is file size how

regards

raj

[Former Member · Accepted solution]

Hi,

Is there any different ways to do splitting the file.

Thsi is a very generic question u asked... what exactly u want?? can u plz explain in detail..

There are many ways to split the file...

if it is record type how or if it is file size how

If record type then u need to split the records in mapping based on its content..

If baseed on size from the source File adapter itself u can split them into many messages.. using

number of records for message..

If still nt clarified,, do post

Babu

[rajasekhar_reddy14]

Hi Raju,

If you want to split the file at the time of picking we caan go with two ways

1)Using Java Proxy we can split the file in to different Chunks.

2)If it is a flat file , if you are suing Sender File Communication channel then use Recordser per message option , there you can give value like 200 or 300,depends on requirement it process the file.

if you want to split at mapping level

1)Go with Java Mapping.

abopve 3 options i implemented in my experiece.

trasfering 50 mb message i think it is not a right idea,you have to split .else you face serious problems like Java stack down,heap memory issues.

Regards,

Raj

[Former Member]

what is max.size xi can proxcess

[Former Member]

Hi,

The Maximum size depends upon many factors including the client infrastrucrure...

So, the thumb rule is always the payload size for PI around 5MB-10MB will give u the best efficiency.

So based on this u restrict the number of records from the source,to get the efficient processing of messages in PI.

Babu

[rajasekhar_reddy14]

There is no specific number from SAP,it is purely depend on the hardware you have and the server configuration

Regards,

Raj

[Former Member]

You can split a file in the following ways: -

1) In the mapping, say you want to split the file after 1000 records, then write a udf to add context change after 1000 records and map that to MT level

2) You can use BPM to split a file based on a condition

[former_member181985]

>if it is record type how or if it is file size how

If it is record type, the mapping we design will take care of things.

if it file (message) size either sender application has to chunk (or) the PI adapter should have this capability.

In general, File Adapter has capability of sending fixed no. of records/message.