Security Architecture, Presented at Cybersecurity Summit

Today’s most prevalent and widely discussed attacks exploit code-level flaws such as buffer overruns and type-invalid input. We need to anticipate tomorrow’s attacks and think beyond buffer overruns, beyond code-level bugs, and beyond the horizon. To be ready for threats of the future, we need to be doing more basic research in cybersecurity today. This talk will outline a few suggestions for important research directions in cybersecurity: the foundations of trustworthy computing, security architectures, privacy, usability, and security metrics.

Most computer security problems arise from buggy code. It seems clear that writing large, bug-free programs is and will remain beyond our abilities. We propose a different goal: protecting what really matters. On e-commerce sites, the web server is primarily a front end for a database. Protecting the latter is much more important than protecting the former. Doing this properly requires a different approach to overall system architecture.