There have recently been a number of very worthwhile posts here at SDN on business semantics and business syntax, and particularly whether business syntax can be extricated from the IT realization of this syntax. (See posts by Rickayzen, Trickonovic, and Reich, for example.) With respect to this "extrication"question, consider the following simple example. Customers want accounting document line item info out of SAP by Profit Center, and they want this info within reasonable response times. OK - the semantics of this request are transparent. But let's consider the syntax (assuming we're NOT in BI with or without BI Accelerator.) Well, we know we have to find our accounting document numbers by going first to GLPCA (because we can't index BSEG by Profit Center unless we de-clusterize it, which is something no one wants to do.) Plus, we know that to be safe, we should take the refdocnr from GLPCA to the awkey of BKPF, even though in a lot of cases we can take the refdocnr from GLPCA directly to the BSEG belnr. But here's where it gets tricky: when we take the refdocnr from GLPCA to the awkey of BKPF, if we want the most efficient indexed read on BKPF, then we have to construct the awkey as: a) number/company/year, if awtyp is BKPF OR b) number/year, if awtyp is MKPF (because MM doesn't know about company numbers). (And note - it's not even number/blank company/year in case (b) - it's just number/year.) So as you can see, there is a huge distance between the business semantics of the original customer need and the details of the relevant business syntax in which these semantics must be expressed/realized. And this distance is, of course, why I have blogged such posts as these in the past: Let's Crystallize the "Transparent Data Model" Discussion: Enterprise SOA, OTC, F-28, and open_fi_p... (and will continue to do so until I'm dragged from the room kicking and screaming in my straitjacket.)
