hi Rambo,
hope this info helps
from 'how to generic delta' doc
https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/docs/library/uuid/84bf4d68-0601-0010-13b5-b06...
Example for Determining Selection Intervals with a Generic Delta:
1. For every delta extraction, there is a selection criteria generated: ?last extraction status
(RSA7)” < x <= Today.
2. Offsets for upper and lower delimiter are subtracted from the respective sides
3. Example 1 shows the normal behavior (both offsets = 0), example 2 shows both safety
intervals (offset for lower delimiter = 2, offset for upper delimiter = 1)
4. Offset for lower delimiter can only be defined for delta method “new status for changed
records
Example 1 (both offsets = 0):
Example 2 (offset = 2 resp. offset = 1):
Tips:
<b>1. If delta field is Date (Record Create Date or change date), then use Upper Limit of 1 day.
This will load Delta in BW as of yesterday. Leave Lower limit blank.</b>
2. If delta field is Time Stamp, then use Upper Limit of equal to 1800 Seconds (30 minutes).
This will load Delta in BW as of 30 minutes old. Leave Lower limit blank.
3. If delta field is a Numeric Pointer i.e. generated record # like in GLPCA table, then use
Lower Limit. Use count 10-100. Leave upper limit blank. If value 10 is used then last 10
records will be loaded again. If a record is created when load was running, those records
may get lost. To prevent this situation, lower limit can be used to backup the starting
sequence number. This may result in some records being processed more than once;
therefore, be sure this DataSources is only feeding an ODS Object
<b>Safety Interval Upper Limit of Delta Selection</b>
This field is used by DataSources that determine their delta generically using a repetitively-increasing field in the extract structure.
The field contains the discrepancy between the current maximum when the delta or delta init extraction took place and the data that has actually been read.
Leaving the value blank increases the risk that the system could not extract records arising during extraction.
Example: A time stamp is used to determine the delta. The time stamp that was last read is 12:00:00. The next delta extraction begins at 12:30:00. In this case, the selection interval is 12:00:00 to 12:30:00. At the end of extraction, the pointer is set to 12:30:00.
A record - for example, a document- is created at 12:25 but not saved until 12:35. It is not contained in the extracted data but, because of its time stamp, is not extracted the next time either.
For this reason, the safety margin between read and transferred data must always be larger than the maximum length of time that it takes to create a record for this DataSource (with a time stamp delta), or it must display an interval that is sufficiently large (for determining delta using a serial number).
<b>Safety Interval Lower Limit</b>
This field contains the value taken from the highest value of the previous delta extraction to determine the lowest value of the time stamp for the next delta extraction.
For example: A time stamp is used to determine a delta. The extracted data is master data: The system only transfers after-images that overwrite the status in the BW. Therefore, a record can be extracted into the BW for such data without any problems.
Taking this into account, the current time stamp can always be used as the upper limit when extracting: The lower limit of the next extraction is not seamlessly joined to the upper limit of the last extraction. Instead, its value is the same as this upper limit minus a safety margin. This safety margin needs to be big enough to contain all values in the extraction which already had a time stamp when the last extraction was carried out but which were not read. Not surprisingly, records can be transferred twice. However, for the reasons above, this is unavoidable.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.