Monday, May 14, 2012
To Cloud or Not To Cloud?
The other day I was working in my lab with a new student, and we ended up with a lot of image data to transfer from his laptop to mine. Because I personally date back to the days when computer data was transferred by means of a stack of paper IBM-type punch cards, my first thought whenever I want to move or store lots of data is to resort to some physical medium: a hard drive or flash (USB) drive, typically. But my student proposed using a service called Dropbox. To use it once it was installed, all he had to do was to put the data in a file on his computer. The software sent it over the Internet to some data center somewhere, and then sent the stuff to be downloaded to my computer where I could access it in a similar file. And it was free, at least for the first two gigabytes of data.
Dropbox is an example of “cloud computing”: the dispersal of computing resources onto the Internet, instead of localizing your computer power in a physical box or boxes at your site. Radioastronomers came up with one of the earliest cloud-computing applications I’m aware of, when they wrote an application to process raw data produced by SETI (the Search for Extraterrestrial Intelligence). If you wanted your computer to help in the SETI search in its spare time, you just downloaded their app and could take comfort in the knowledge that you were one of hundreds of people all over the country helping SETI look for extraterrestrials.
Nowadays, of course, cloud computing is a big deal business-wise, as companies recognize that outsourcing a lot of their IT needs makes more sense than trying to maintain their own physical system with all the hassles that involves. But I wasn’t aware of the ethical implications of cloud computing until I came across an article by Dunstan A. Hope and Ryan Schuchard on greenbiz.com, an Internet publication for businesses interested in being more environmentally conscious.
It turns out that the “cloud” is, of course, no airy nothing floating around in the ether, but consists of servers, processors, power supplies, cooling systems, and (a few) maintenance personnel concentrated in “data centers” whose locations are not always public knowledge. It’s understandable for security reasons that companies who run these centers aren’t just posting their addresses everywhere, but their geographic anonymity makes it easy to assume that the cloud really is a cloud, and has no needs for space, electricity, water, or other resources. It’s a little like things were back before we started being environmentally conscious in general: when you threw something away in those innocent times, you didn’t give a second thought to where “away” was. But now we know better, or at least we should.
Large data centers run by outfits such as Google use so much power that they are located near sources of abundant cheap energy. One estimate by the Environmental Protection Agency says as much as 1.5% of the U. S. electric power output is used by data centers. This can be hydroelectric energy, such as The Dalles, Oregon’s Columbia River power, or coal-fired power plants in the Midwest. It’s an open question, as far as I know, whether it’s more energy-efficient for 100 businesses to use the cloud-computing services of one data-center operation, or for them all to have their own computers in their own locations. I suspect if the data center is run with an eye toward energy efficiency, it may be better energy-wise to use the cloud. A new trend in data centers is to build them in arctic areas so that you can use natural cooling (basically running with the windows open, so to speak) even in the summer, rather than pay for expensive refrigeration machinery to cool the systems in hot weather. But there are not that many arctic areas with abundant cheap energy, so there are problems with this idea too.
Besides the notion of energy conservation, there is the question of security. I confess to an atavistic feeling that the best measure of security for my data is if I can hold its physical embodiment in my hands: a flash drive, a hard drive, or a laptop where the data is physically stored. But realistically, a better way to protect against data loss is to hand it to professionals who put it on multiply-backed-up remote servers such as the Dropbox people or many other Internet services provide. I suppose some malevolent malware-writer could cause a wipeout of the data stored in an entire cloud-computing service’s files, but it would be hard, and not nearly as likely as a hard-drive crash on one individual’s computer. I always keep backups, but backups can fail too, and there’s the bother of keeping track of the media, updating it as it goes to legacy status, and so on. So cloud computing makes sense from a data-security standpoint.
Besides physical security, there is the question of somebody stealing data or otherwise gaining unauthorized access to it. The banks have dealt with this type of problem since the first bank began using the first computer, and while Americans are notoriously sensitive about breaches of their personal financial data, nobody much seems bothered by the fact that your personal financial information is stored in scattered places around the country. Of course, not all cloud-computing firms have security as good as bank data systems, but at least the precedent is there. So I’m not so concerned about this aspect of cloud computing.
Whatever the ethics of the trend, it looks like cloud computing will be in our future more as time goes on. If you use a cloud-computing service, you can make an effort to find out what their Power Usage Effectiveness (PUE) rating is. This is the ratio of the total power used by the facility divided by the power actually needed by the computing equipment. A lower number (lower than 2) is better. And if they provide such information, find out where their data servers are, and what kind of power they use. Even if it’s billed as a free service, somebody’s paying for electricity somewhere, and you might as well be responsible enough to find out about it.
Sources: The article “Cloud computing raises new ethics, sustainability issues” appeared at http://www.greenbiz.com/blog/2012/04/11/ethics-cloud-computing?page=0%2C0 and was written by Dunstan Allison Hope and Ryan Schuchard. I referred to the articles on data centers and The Dalles on Wikipedia. And I use Google’s cloud-computing service blogspot.com to post this blog, although I always keep a copy on my laptop!