I have seen recently a lot of problems or SCN discussions under BW area about the use of Error Handling and Semantic groups on DTP’s.


And I thought, it would be a good idea to give a brief overview of the use of these features on DTP’s.


The goal of this blog post is provide a generic information about the Influence of START/END Routine in Transformation on the Processing mode of Data Transfer Process (DTP) for loading Datastore Object (DSO) and the technical reason behind it. In all the cases it is assumed that either a START routine or END routine or both are used in the transformation connecting the source and DSO which is the target. The cases are broadly described below:


  • A1: Semantic Group is not defined in the DTP, Parallel Extraction flag is checked and Error handling is switched off ie. either ‘Deactivated’ or set to ‘No update No Reporting’: Processing Mode of the DTP is ‘Parallel Extraction and Processing’.


  • A2: Semantic Group is not defined in the DTP, Parallel Extraction flag is not checked and Error handling is switched off ie. either ‘Deactivated’ or set to ‘No update No reporting’:Processing mode of the DTP is ‘Serial Extraction, Immediate Parallel Processing’.


  • A3: Semantic Group is not defined in the DTP, Error handling is switched on ie. either ‘Valid Records Update, No reporting (Request Red) or ‘Valid Records Update, Reporting Possible (Request Green)’: Processing Mode of the DTP is ‘Serial Extraction and Processing of Source Package’. The system also prompts a message ‘Use of Semantic Grouping’.


  • B1: Semantic Group is defined in the DTP and Error Handling is switched off ie. either ‘Deactivated’ or set to ‘No update No Reporting’: Processing Mode of the DTP: ‘Serial Extraction, Immediate Parallel Processing’. The system also prompts a message ‘If possible dont use semantic grouping’


  • B2: Semantic Group is defined in the DTP and Error Handling is switched on ie. either ‘Valid Records Update, No reporting (Request Red) or ‘Valid Records Update, Reporting Possible (Request Green)’: Processing Mode of the DTP is ‘Serial Extraction, Immediate Parallel Processing’.



In any DSO we allow the aggregation ‘OVERWRITE’ along with ‘MAX’, ‘MIN’ and ‘SUM’ which is non cumulative. So it is very important that chronological sequence of the records are intact for update because the ‘principle of last wins’ needs to be maintained. Therefore if Error handling is switched on and there are errors in the update then the erreneous records which are filtered out and written in the errorstack must be in chronological sequence.


The solution for the cases described above are:


  • In cases A1 and A2 the error handling is switched off so if there is one error the load is terminated and erreneous records are not stored anywhere. Therefore according to the the setting ‘Parallel Extraction’ ie. whether is checked or not the processing mode of the DTP is ‘Parallel Extraction and Processing’ or ‘Serial Extraction, Immediate Parallel Processing’ respectively.


  • In case of B1 you have defined a Semantic Group which ensures that the records pertianing to the same keys defined in the semantic group are ensured to be in one package. But as error handling is switched off this contradicts the setting of the semantic group as you dont want erroroneous records to written in the errorstack and so the processing mode of the DTP is ‘Serial Extraction, Immediate Parallel Processing’. Also the system prompts to remove the Semantic group otherwise it makes no sense.


  • In case of A3 the semantic group is not defined but the errorhandling is switched on so it means that erreneous records are needed to be written in the errorstack and the chronological sequence needs to be maintained. But as there are no keys in semantic group defined it cannot be ensured that records pertaining to the same keys are in the same package. So the processing mode of the DTP is ‘Serial Extraction and Processing of the source package’.Also the system prompts you to use the semantic groups sothat it is ensured that the records pertaining to the same keys defined in the semantic group are ensured to be in same package.


  • In case of B2 semantic group is also defined and errorhandling is switched on. This means the records pertaining to the same keys defined in the semantic group are ensured to be in one package and if the errors happen then the chronogical sequence will be maintained while writing the erreneous records in the errorstack after they are sorted according to the keys. So the processing mode of the DTP is ‘Serial Extraction, Immediate Parallel Processing’.


I hope this blog can help you in further questions about use of error handling, semantic groups on DTP’s.


Regards,

Janaina


To report this post you need to login first.

1 Comment

You must be Logged on to comment or reply to a post.

  1. KRISHNA YALAMANCHI

    Janaina Steffens Great blog!! 2 Pictures to explain the same 🙂

    With semantic Key – If multiple updates happen to the same record they are all grouped togather in the same package. Takes a little longer to run since it has to sort all the data before it is split into multiple packages.

    /wp-content/uploads/2015/09/no_semantic_key_784243.jpg

    /wp-content/uploads/2015/09/with_semantic_key_784283.jpg

    (0) 

Leave a Reply