Add a Comment Now - We Want to Hear From You
By Brian Boyko
Editor, Network Performance Daily
The Austin Chronicle - Austin's local alternative weekly newspaper - reported today that police busted a pot-growing operation in a rental property based on "data-mining" that they did using Austin Energy's customer database.
Colby's lawyer, David Dudley, argues that APD Detective Jeff Haynes used energy-consumption information from thousands of Austin Energy customers - without those customers' knowledge or consent - in an effort to find and focus upon AE customers who, in Haynes' opinion, regularly consume more kilowatt-hours than he believed "typical" for the size of their residence. In other words, rather than developing an investigation and then accessing a discrete and particular set of energy-consumption data in order to verify other evidence that suggests a growing operation might be under way inside a particular house, Haynes was working with AE, and in particular with business process analyst Mark Coffey, to troll through thousands of records from across Austin in an effort to find energy-consumption "targets" to pursue.
AE officials insist their cooperation with APD is required of the city utility - pursuant to a 1994 ruling from the Texas attorney general's office and supported by federal legal precedent.
Now, I'm doing some thinking - At any particular time, in my smallish apartment, which I share with two other twenty-something geeks, we've got the media center PC going, three computers - sometimes four, if my laptop is booted up. And since we're gamers, those power supplies suck up a lot of energy. It's possible that if police are choosing who to make as a target of investigation a function of a kilowatt-hours consumed divided by square footage of real estate, there are a large number of geeks who live in $400/mo, small apartments so that they can afford their 1000W gaming-rig power supplies.
Hitting closer to home, this could hit those networking engineers that run networks of computers at home - as a way to try out new ideas before they hit the mission-critical servers, or programmers who build their own Beowulf clusters at home so that programs will compile faster. That's not including our HDTVs, our game consoles, etc.
One of the hallmarks of the technically literate is that we like to use technology. And, of course, our technology is based primarily on electricity.
So there's the very real chance of false positives if this method of determining which households are secretly pot greenhouses. Ignoring the privacy implications for a second, the other part of this seems to be a logical trap - If A implies B, then B implies A - which can often stymie those who look for problems in their enterprise networks - the equivalent of saying that copyright infringers use high bandwidth, therefore those who use high bandwidth must be copyright infringing.
No software nor appliances will correct a mistake in reasoning and logical deduction - which is why network engineers have to remain open-minded to problems and refrain from jumping to conclusions when diagnosing network problems.
With the caveat that what I'm suggesting would be a huge expense with returns on investment that might not be apparent for decades…
I wonder about some of the logical conclusions to the trends we're seeing in IT. Specifically, we can now replace inefficient legacy servers with virtual machines on processors that require less energy than the previous generation - consolidating servers. At the same time, WAN optimization allows us to put all those servers in one place, instead of having multiple branch-office servers. These are the two most important current trends in networking right now, and they lead me to a simple thought:
If we need fewer computers, and distance is no longer as daunting as it was, why not vary the location of the servers so that there's an additional benefit.
By all means, you want to locate a company’s headquarters in the area where it makes the most economic sense to do so – but the data center is and always will be a cost of doing business and not a direct revenue generator itself, for most companies. As such, it almost doesn’t matter where the data center is, so long as the data still comes through reliably.
So it might make some sense to consider where the computers would cause the least amount of recurring cost. And one of the big recurring costs is power consumption - not just for the computers themselves, but for the air conditioning required to cool the computers.
Wouldn't it make sense, then, to put the data center in a colder climate, so less air conditioning is needed to cool it, fewer generators need to be purchased, and cooling consumes less energy?
Of course, there are tons of problems with this idea - the additional costs of heating the offices of the staff that must be located in the datacenter facilities - unless you have a plan for dealing with the static electricity discharges generated by parkas, and are willing to train the Siberian Husky teams the IT group will use to get from server to server not to mark their territory on anything mission critical.
In fact, it is very common for data centers to be remote from HQ. This is done to tap into affordable labor markets, provide a remote fail-over/disaster recovery location, reduce the risk of sabotage/terrorism, and avoid natural disasters such as earthquakes, tornadoes, hurricanes in South FL, and yes, blizzards.
However, IMO the “green” issue is going to become big news in the US over the next couple of years, following Europe’s lead. C-level executives will want to tout their initiatives to reduce data center power consumption through new techniques.
Just a little thought exercise. So, readers, are you seeing anything like this happening in your own organizations?
