Andrew D Wolfe Jr, Boston University
In data modeling, many of us focus rightly on normalization, functional dependencies, and tuning. The recurring crises in information security, it seems, pass us right by. We focus on data semantics and on projecting business constructs into relational databases. We make minimal use of DBMS security features, and then blame breaches on the network engineers or the application programmers. This, however, is like blaming a bank’s exterior walls after bank robbers penetrate, then find money not in a vault, but stacked on open, unprotected shelving.
The InfoSec precept of defense in depth must go within the firewall, within the network zones, within the application code, within the persistence framework and into the database itself. But how do we make our business- and data- focused models embody data security as well? By recognizing that the authorization of data access is itself a key business process. Users, data assets, and data ownership are how the organization works, not some extrinsic unpleasantness. We present a way to integrate InfoSec principles during the entire data modeling (and re-modeling) process, along with selected hands-on exercises. Armed with these techniques, we can finally capitalize on strong DBMS security features, posing would-be data robbers who penetrate our outer bulwarks with a database design that presents the hardest defense of all.
Andrew D. Wolfe, Jr., M.S.C.S, brings over thirty years’ experience in designing and delivering database applications to the graduate Computer Science students he teaches at Boston University’s Metropolitan College. Mr. Wolfe specializes in data modeling, database information, and information security.