Skip to Content
avatar image
Former Member

Create a custom table in PI ABAP Stack

Is there a performance or impact to Integration Server, if users are allowed to maintain entries in custom tables in PI database.

Specifically I request best practices of allowing access to PI database, other than internal SAP PI services. I am concerned that if any issues wiith database occurs, it may impact the whole Integration server and impact the Java Stack and bring down the system. I am leaning towards creating external database and access using JDBC to minimize the risks to integration server.

Appreciate any help or suggestions.

Regards

PM

Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

2 Answers

  • Nov 10, 2009 at 11:21 PM

    it may impact the whole Integration server and impact the Java Stack and bring down the system

    Not really. It won't be that big a problem unless you mess with the standard tables. The custom tables on ABAP stack would no way affect the Java stack.

    As for the best practice, it is obviously better to maintain the tables on an external database. But usually if there is no huge data involved, you may use tables on PI system and use ABAP mapping to access them.

    Regards,

    Prateek

    Add comment
    10|10000 characters needed characters exceeded

  • avatar image
    Former Member
    Nov 11, 2009 at 09:27 AM

    Hi,

    I guess in connection with ABAP mapping the use of customer tables in the PI database is a good approach and you won't see any performance issues (of course, you should check with your PI administrator, if the tablespaces of the database are big enough).

    In an implementation project, we had a "SAP idoc -> PI -> Oracle DB (JDBC) "-scenario (ORDERSP-idoc). In the Oracle Database we had to distinguish if it' a ew order or if it's a change to an existing order. This was performed via ABAP-mapping in the following way:

    First we created a customer table on PI. Table fields were document number (key) and a character type field to store a checksum. The ABAP mapping performed the following sequence:

    1. calculation of a checksum with the idoc-xml structure (of course we had to exclude document number etc. from the string representation of the xml.)

    2. check if the document number is already stored in a customer table.

    3. if no, we had a new record, if yes ->

    3a. comparsion of the checksum with the checksum stored in an customer table on PI (key was document number). If the checksum was different, we had a change-record, otherwise we had to ignore the whole idoc.

    4. save the document number and checksum in the customer table.

    The scenario is working very well and we had never observed any performance issues on PI.

    Best regards,

    Herwig

    Add comment
    10|10000 characters needed characters exceeded