Snippets

It never hurts to take a snapshot of high-profile tables in the production system before making changes. This task can be a manual task that is performed each day or broken down into object-specific tasks that automatically export pertinent tables as part of the Convert Data step.

SAP expects date and number information to be formatted a certain way, but it varies based upon local standards and is configured for each client. When writing software and testing in a development system the software depends on the localized standards in the programmer's User Profile. The User Profile can be changed using transaction SU3 or by clicking through the menu with System → User Profile → Own Data.

When performing pre-validations on data and comparing it against existing data the job tends to run a long time. LSMW by default will indicate the progress of the conversion step after every 500 records, but it doesn't indicate how much longer the task will run. While unnecessary, it is a fun way to plan for coffee breaks.

SAPGUI 7.10 and earlier
SAPGUI 7.20 and after

When assigning classifications to materials it helps to check to see if the material is already assigned to a classification. In some cases materials can have just one type of classification assigned to it. For example, materials cannot have multiple batch classifications assigned to them.

Sometimes problems in source code are too complicated to resolve without stepping through the code while it is executing. Adding a break point to the ABAP code in LSMW is easy and provides an invaluable way to debug complicated problems and is faster than invoking the debugger with /h.

Often the behaviour of an LSMW object needs to change quickly to fit requirements as a project progresses. Modifying the source code in a development system and then transporting the LMSW object to a quality or production system takes time. Writing multiple LSMW objects for slightly different tasks is cumbersome and hard-to-manage and it duplicates effort.

Excel has a lot of tendencies that make it a less-than-desirable tool for data conversion. Too often data is corrupted by Excel's tendency to convert numbers from one international standard to another, to convert long numeric values to scientific notation and to convert dates from one format to another. The insufferable habit of trimming off leading zeros makes Excel near worthless for handling many types of SAP data.

Where Excel shines is in it's ability to act as an easy-to-use reporting tool. Handing off reports for the purposes of validation and reconciliation is best done in Excel because few functional resources are equipped to handle raw text. Additionally, Excel is the industry standard for perusing table data and is installed on almost all work computers.

Many legacy systems have little or no validation on the postal code and SAP has very strict, country-specific validation that cannot be circumvented. Validating the postal codes in LSMW as part of the Convert Data phase is always a good idea and is a great way to identify defects in the data and provide feedback to the client. Early in the implementation it is better to transport the master data with dummy postal codes to allow the processing of downstream transactional data.

So often a disconnect appears between the Vendor Masters that are transported to SAP and the transactional data that follows. Vendors are a dependency for documents like accounts payable and purchase orders. If vendors aren't present in the new system or, if they're configured incorrectly, then the documents cannot post and a conversion project can grind to a standstill.

All versions of MS Access have a 2GB file-size limitation. Without warning or explanation, Access will simply stop working and the database is rendered useless until the size is reduced. Handling large text file extracts from legacy systems as linked tables is an easy workaround for this problem, but managing linked tables is cumbersome unless every single revision of the extract is in the exact same format.

Linking to the text file rather than importing it presents a non-negligible performance hit. With modern computers linking doesn't have a significant impact on overall processing time, especially if the text file is on a local hard drive instead of a network share.