Threats are divided into broad categories, including insiders, script kiddies, nation-states, terrorist groups and forces of nature, among others. Threat capabilities are then considered, with assigned scores ranging from 0 (no known capability) to 5 (no threat more capable). Areas of capabilities are also applied, such as an attacker's institutional knowledge, technical proficiency, group size and funding, and levels of access.

"The benefits are not only immediate and enable us to refine our prioritization of remediation activities, but now we're beginning to see a huge advantage in longer-term trending," Bromberger says. "Last year, we may have been worried about five particular threat agents; this year, it's five others. This enables us to more precisely target the implementation of our compensating controls."

Gene Schultz, CTO of High Tower Software, has advised Bromberger on the project. He says most threat modeling is more theoretical and academic. "What makes what Seth is doing so...

If there are points to challenge with this methodology, you could start with the fact that the intelligence gathered on threats has a shelf life and must be updated regularly. Also, you must consider the weekly labor-intensive demands of processing hundreds of pieces of updated vulnerability intelligence.

"It's difficult; this is only as good as the qualitative data you feed it," Bromberger says.

Bromberger says, with management's continued support, PG&E may publish the methodology once it has been subjected to more peer review.

"One advantage to this methodology is the cost of development and implementation; this is labor. It's about brainstorming and bringing people together," he says. "It's just a matter of rigor, and if it demonstrates a level of security that gets you a 3 on IA-CMM, I'd say that's worth it."

< PREV PAGE   |   1  |   2  |   NEXT PAGE  >