As probably many of us, I have been tasked to model the data that the company maintains in SAP. But SAP contains tenths of thousands of tables, most of which not even understandable, and does not provide - at least to my knowledge - any valuable documentation on how the data are "organised".
Does anyone have a suggestion on how to model data in SAP?
Steve Hoberman alerted me to your post.
I can sympathise with your challenge and think we can help you. Trying to model SAP data manually is extremely difficult. As you have probably figured out there are no descriptions for tables and attributes in the database system catalog. In addition you cannot find how tables are joined from the system catalog. Any documentation you can find is unlikely to be of much help if your SAP system has been customised.
All the information you need for that and more is found in the SAP data dictionary tables. When you access those you then have the problem of trying to organise all the metadata so that you can navigate and make sense of it for your model.
I would also suggest that it is not practical to make a model with tens of thousands of tables. You are likely to find that many thousands of your tables have no data or are not relevant to you anyway.
I work for a company called Silwood Technology. We make a product called Safyr which is designed to make it quick and easy for you to model your SAP system and share the results for onward processing in data modeling tools and elsewhere. Safyr is SAP certified. Safyr is used by many SAP customers to help them model their SAP data.
Safyr connects to the data dictionary tables and extracts the rich metadata they hold including customisations and stores it in a repository. You can then navigate, search, filter and analyse the metadata in order to create subsets (Safyr Subject Areas) which are relevant to your needs e.g. to represent a specific business topic. Finally those Subject Areas can be shared with data modeling tools (ER/Studio, erwin, SPA PowerDesigner), data catalogs (such as Collibra, Alation, Informatica EDC) and in other formats for further work.
I have attached a pdf with 3 images which I hope give you an idea about how you could use Safyr.
If you would like to know more or see a demonstration please contact me on the email address below and we can set up a call.
This can be done, but only very slowly and with much input from the users.
I reverse engineered SAP A&D solution at a major international airline, and it took the best part of six months to come up with a decent structure of data for BI/Warehousing purposes, which was only 20% of the tables present in the schema. Having access to the metadata will get a long way, but it often becomes a process of elimination, removing the tables you know are not relevant, and assuming the rest are the ones that contain business data. Even then, user input is utterly invaluable to understanding the relevant size and scale of the tables that contain the data that really matters.