Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.
Originally posted by: Tim MeagherI've updated the library attached to the first post again.
It is still compatible with LAL 2.18.
The changes have been made to significantly improve performance on the Extract ERP Table node when extracting data from tables who's fields are larger than can be populated in the RFC_READ_TABLE call (default 512 bytes).
Previously, the node would need to process small subsets of the data, executing the RFC_READ_TABLE function to extract each of the parts of the table before loading them into memory.
Since the table data was loaded into memory, the number of records per extract needed to be small, resulting in many function calls.
This no longer needs to be done.
As such, the RowsPerExtract parameter has been removed from the node.
Initial testing indicated a significant performance improvement, whereby extracting data from one of the tables (PA0006) on a SAP test system was previously taking 12 minutes, while it now takes 27 seconds on the same system.
During testing, also investigated further the effect of the RowBatchSize parameter.
By default this is 10,000.
Increasing this to about 100,000 displayed significant performance improvements as well.
Therefore, for optimal performance, this parameter can also be tweaked.
However, setting this parameter too high will result in the SAP JCO code storing more records than it can allocate in memory, so care must be taken with this parameter to ensure that it doesn't exceed the memory restrictions on the node or of the system on which the node is running.
Tim.