DHS To Review Report On US Power Grid Vulnerability

CWmike writes "The US Department of Homeland Security is looking at a report by a research scientist in China that shows how a well-placed attack against a small power subnetwork could trigger a cascading failure of the entire West Coast power grid. Jian-Wei Wang, a network analyst at China's Dalian University of Technology, used publicly available information to model how the West Coast grid and its component subnetworks are connected. Wang and another colleague then investigated how a major outage in one subnetwork would affect adjacent subnetworks. New Scientist magazine reported on this a week or so ago, and the paper has been available since the spring."

Comments

Don't worry

[dedazo]

The US power grid is so ancient, convoluted and in such a massive state of disrepair that we can be sure we're safe from terrorists. They wouldn't even know where to begin to find a point in the system that could be used to trigger a catastrophic cascading failure like the one in the East Coast a few years ago.

Trees on the other hand... trees are truly evil.

Re:Don't worry

[Saliegh]

Everyone knows that a small thermal exhaust port at the end of a long trench is the key to initiating a catastrophic cascading failure.

The amazing thing

[guruevi]

The amazing thing is that nobody ever tried it or at least never succeeded. The US is apparently not that hated in the world since nobody ever does anything. We have hundreds of reports on how easy it would be to disable this or take that out of service. All it takes to black out the USA are some well placed charges or for somebody to hit a few poles hard enough but nobody does it. All we got was some measly hijacked plane (which has been done since the 70's) in a few buildings.

Re:The amazing thing

[hoggoth]

> It didn't topple our economy, which was their aim, but put a dent in it.

Yes it did. The cost of the buildings is negligible compared to our GNP. But the cost of the followup war in Afghanistan, war in Iraq, DHS, etc have toppled our already shaky economy. What's more their aim wasn't to topple our economy, it was to ruin our way of life. I'd say our descent into security theater, torture, surveillance and paranoia has gone a long way towards destroying our way of life. America the free?

Re:Power Station PLCs should _not_ be connected...

[AB3A]

Not so fast. See the first paper in this bunch. The authors managed to hack a Koyo and AB PLC Ethernet interfaces. The AB Ethernet card had lots of useful stuff in it, including a symbol table. From the symbol table I saw many backplane calls that you could use to communicate with the PLC. How well do you trust a hacked Ethernet module on a PLC backplane?

Having a physically separate port is nice, but it is no substitute for secure coding. If you think that coding is poorly secured in the PC world, you'll be shocked at what often gets done in embedded system coding.

Some PLCs and Variable Frequency Drives have been noted for their inability to handle Denial of Service traffic. I've seen that demonstrated myself. This is the official cause of a reactor SCRAM at Browns Ferry a few years ago.

Try a port scan of your PLC some time and tell me how many ports it responds to (DO THIS ON A TEST-BENCH --NOT PRODUCTION EQUIPMENT!). If you can identify everything that critter responds to, congratulations. If not, be afraid. Be VERY afraid. I've heard quite a few PLC models that have mysterious responses to ports where you wouldn't expect them to respond.

Real Time embedded systems are not good candidates for direct internet exposure. They're too difficult to patch in a timely fashion. Often the windshield time alone is prohibitive. And if you have any notions of pushing patches to them remotely, remember, these things control some pretty high speed/high power processes. You don't just patch them. There are process and safety implications that you need to consider. This ain't some office application where you can say oops and restore from a backup. Real physical things will happen and real physical problems will be created that you can't clean up with a simple code reversion.

Most of our infrastructure today has not been engineered with security issues in mind. There is still lots of Gee Whiz "Let's Share Data" synergy crap going on. This leads to all sorts of direct interconnections that aren't absolutely necessary. Many controls can be made over links that weren't intended for that purpose. It's not easy to split the data flows up any more because many organizations have been very profligate with their use of SCADA information and it isn't easy to find all the sources and sinks.

I'd love to post data from a PLC directly to the public. But I just can't sleep at night with something like that waiting to screw things up.

Good luck with your security, and I mean that quite sincerely.