Over the next week in Vegas, Jonathan Pollet and I will be making our annual trip to the desert, to teach our SCADA security course out at UBM’s Blackhat Training & Briefing event. In response to increased media-created fear, uncertainty and doubt over whether groups like Anonymous are capable of successfully attacking the electric grid, through its various digital interconnections, we’ll be spending a half day discussing what exactly would be required / involved in pulling off a successful attack against the US bulk electric system (or “BES”). The point of this activity is not so much to provide attendees with a road-map to conducting such an attack (indeed, this couldn’t be learned in a half day), but to understand the relative complexity of pulling off such an attack, owing to the highly heterogeneous nature of the target environment, therefore making it fairly laughable that anyone except a well-funded nation state could likely pull off a broad-scale attack. For those of you who won’t be able to join us out in LV, here’s the skinny for your enjoyment and hopefully, to provide you some talking points next time you’re asked about such craziness.
One, the electric grid isn’t a single grid. In Northern America (and Europe), the “grid” is actually comprised of multiple, regionalized grids, each with its own ability to sectionalize and isolate itself from outages which may be occurring inside and outside of its operating region. The two primary grids that you’ll hear people talk about are the western and eastern interconnections; however, Texas also has its own grid, as does Quebec (although generally considered to be part of the eastern US interconnection). The point here is that you’re not just targeting one grid, you’re at very least targeting four, each with their own fault isolation and monitoring systems which would need subverting.
Two, the electric grid doesn’t run Windows. Well, at least not most of it. If you consider the numbers of “assets” involved in the operation of the grid, a very small percentile are running any kind of standardized operating platform, and where they do, mostly exist in control / monitoring rooms. When we’re talking about doing physical damage to power lines, transformers and generators, we’re now talking about an extremely heterogeneous environment, which you will find seldom configured in any kind of symmetrical manner, even within a given electric utility. This means that in order to attack this type of equipment , you n not only need to know make, model, firmware version etc, but you also need to understand the context under which the device has been deployed in order to effect the desired impact on whatever it happens to be hooked up to. Even if you’re able to compromise a more monolithic layer, such as a control room, this will only buy you an interface to the field-equipment (PLC’s RTU’s etc) which is what ultimately needs manipulating to make things go bang and lights go out. It is this nuance which made Stuxnet so sophisticated, given its ability to target highly specific industrial processes and the field devices (PLC’s), which were associated with nuclear enrichment.
Finally, it is not to say that such an attack would be impossible, it is just unlikely to manifest itself in the way that we generally view the smash-and-grab nature of most cyber-attacks. The military doctrine of preparing the battlefield is much more likely to apply, which is to say; that a well-funded, long-term campaign would likely be necessary, to gather sufficient intelligence on the environment and position sufficient resources, such that once critical mass is reached, an event may be triggered (such as a cascading failure within each grid geography) which would overcome countermeasures in place. This isn’t something that can happen overnight or come cheaply.