Slashdot Mirror
Power Grid Insecurities Examined
Joe Barr writes "Chris Gulker has taken a long and careful look at the infrastructure of our power grids and has come to some rather unsettling conclusions." A good read that outlines where the current power grid is at, and suggests some paths for the future that may help avoid future blackouts.
Says Skroch: "If you have too much security [i.e., no network connections], then the power plant probably won't work."
power plants worked long before the internet was created. no important computer controlling very important things should ever be put on the internet.
The power industry needs to be reinvesting profits in infrastructure (powerlines), not stock dividends. The same companies should have been upgrading their command and control systems to prevent chain reaction blackouts. Am I expected to believe the computer systems that manage the cooling rods in the nearest nuke plant are secure?
Seriously consider the economic impact of the grid failure compared to the recent worm problems. Then think about a nasty combination of the two.
Karma: Censored (mostly affected by decency laws)
We're starting to see a few problems appear more than once, though.
-
Telecom vulnerability to power failure.
-
"Non-critical" systems that aren't.
-
Cross-connection between business systems and control systems
That's a more realistic picture of what's going on.AT&T was determinedly independent of the power grid in the days of Ma Bell. Every central office ran on 48VDC storage batteries, with backup generators. The backup generators were started once a week, and run for several hours once a month. Once a year, each central office ran for 24 hours cut off from external power.
That was a long time ago, back when AT&T was a regulated monopoly common carrier. In the new, competitive era, that depth of backup can no longer be assumed. Carriers in trouble (WorldCom, Adelphia) tend to cut things like that.
The details aren't in yet, but it's beginning to look as if, during the recent big blackout, some comm links went down very early, so that the fault information that's supposed to divide the grid cleanly into islands didn't get through. Once all the logs have been correlated, it will be clear what happened.
A few weeks ago, CSX, the railroad, had a shutdown due to a virus. Railroad signalling has used "code lines" for decades, for remote control of switches and signals. These are basically serial links over which commands and responses are sent. The safety logic is local, but if you lose a code line, the dispatcher can't throw switches and route trains.
The tendency to centralize train control has resulted in a need to transmit code line signals hundreds or thousands of miles. So they tend to be multiplexed over telecom-like facilities. CSX apparently routed theirs over their in-house general purpose network. The routers in that network were managed by a network management system that ran on Windows. When the Windows machines went down, system management of the routers stopped, and, after a while, this apparently took some key routers down. So a "non-critical" system actually stopped train movements.
It's really convenient to be able to see what the plant is doing from your desktop. Order processing is more efficient if the sales network connects to the factory network. Energy traders need to be able to see what the power plants are doing, and give directions to power dispatchers. These things all create vulnerable paths.