Nick Carr: The ways cloud computing will disrupt IT
- 26 March, 2009 08:35
Whether you prefer the term "utility computing" or "the cloud," the industry is headed in that direction, however slowly, and the transition will have a multifaceted impact on IT in some ways productive, others unpleasant. And it will strike to the heart of the very technology professionals who provide a significant chunk of what is today's enterprise IT.
Nicholas Carr, author of the tech-contentious Harvard Business Review article "IT Doesn't Matter" and, more recently a book "The Big Switch," spoke with Tom Sullivan about how enterprises will transition to a more utility-like model for IT, why a small cadre of companies is gobbling up 20 percent of the world's servers and the unheard of possibilities that creates, how Web 2.0 replicates business fundamentals, as well the human factor in all of this.
There are those who would say that The Big Switch is something of a shift away from your position in "IT Doesn't Matter." Is that true?
Even as far back as my original HBR article " IT Doesn't Matter," one of the basic arguments was that more and more of the IT that companies are investing in and running themselves looks a lot like infrastructure that doesn't give you a competitive advantage. I even made an analogy with a utility that everybody has to use but becomes over time pretty much a shared infrastructure. And so what happens when most of what companies use is indistinguishable for what their competitors use? Doesn't that mean we'll move toward more a shared infrastructure, more of a utility system?
The rise of cloud computing in general reflects the fact that a whole lot of the IT that companies have been investing in is really better run centrally and shared by a bunch of companies than it is maintained individually. Now, having said that, I see it as a logical next step from "IT Doesn't Matter." On the other hand, you could say that if a company is smarter in how it takes advantage of this new technological phenomenon, it might at least get a cost advantage over its competitors. So at that level, you can say there's some tension between the idea that you can't get an advantage from technology and what we're seeing now with cloud computing.
A shift toward IT as a utility will be long-term. It's happening in really small ways, such as Salesforce.com, but in the here-and-now, not a lot is going on. So do you have a timeline for this?
It's a 10- to 15-year period of transition. Particularly if you look at large companies, big enterprise users, that's probably about the right timeframe. They have huge scale in their internal operations, and it's going to take quite some time for the utility infrastructure to get big enough and efficient enough to provide an alternative to big private datacenters. I completely agree that we're in a transition period that's going to take some time. And I'd say at this moment that the hype about cloud computing has gotten a bit ahead of the reality. Still, the uptake of cloud services over the last year has moved faster than I would have expected. So there is a lot going on, but there's still a long way to go.
Page BreakIndeed, and there's also a human element at play. IT folks are in the position to push back on cloud services, if only to preserve their own jobs. How do you envision that playing out?
Well, there is a basic conflict of interest that IT departments face as they think about the cloud, and that's true, of course, of any kind of internal department that faces the prospect of being displaced by an outside provider. We also saw some of this with outsourcing as well, so it's not necessarily new. What I think is more powerful than the resistance that may come from IT departments looking to protect their turf is the competitive necessity companies face to reduce the cost of IT while simultaneously expanding their IT capacity -- and the cloud offers one good way to do that. What I mean by competitive pressure is if one of your competitors moves to more of a cloud operation and saves a lot of money, then whether your IT department likes it or not, you're going to have a competitive necessity to move in that same direction. Over time, that is going to be the dynamic of why cloud computing becomes more mainstream.
And that leads to fewer IT folks, even though some say cloud services won't eliminate IT jobs?
NC: You hear that all the time, not only from vendors but IT managers as well, who say, "If we can get rid of these responsibilities, then we'll be able to redeploy our staff for more strategic purposes." That's a myth, obviously. Any time you get rid of a job, then the person in that job has to prove that their continued employment is worth it for the company. When you get rid of a job, more often than not, the employee goes out the door, particularly when you have an economy like today in which companies are looking not only for greater efficiencies but also to reduce their staff.
How do you see the current recession reshaping IT?
I have mixed feelings. On the one hand, it continues or even intensifies the cost pressures that have been on CIOs and IT departments over the last decade, and that would seem to imply there will be a search for more efficient and less costly ways to do the things you need to keep your business running, whether that's purely at the level of computing and storage capacity, or how you get a particular application in. From that standpoint, it should promote the use of cloud services simply because at the outset they're much cheaper and don't require capital outlays.
On the other hand, whenever you have the kind of severe economic downturn we have right now, companies tend to get very conservative and very risk-averse and so they might be less willing to experiment with a new model of IT. So there are these two contrary forces at work: One pushing companies to find more efficient ways to do things, and the other kind of this sense that "let's batten down the hatches and not do any experiments." And I don't know how those two forces play out. Earlier signs, if you look at Salesforce's results, which have held up pretty well so far, would indicate that maybe it is pushing companies to move more quickly to explore or even buy into the cloud, but it's too early to make a definitive statement on that.
Page BreakI read your recent blog post in which you state, in short, that a handful of big companies -- Microsoft, Yahoo, Google, and Amazon -- are buying about 20 percent of the servers sold today, and that fact along with more powerful edge devices, is leading to a new architecture that makes applications unheard of before now possible. What are some examples?
I wish I knew. I don't mean that in a glib way. We don't know yet. The New York Times' use of Amazon.com is a small but telling example of what happens when you radically democratize computing so that anyone has access at any moment to supercomputer-type capacity and all the data storage they need. So it's about what can you do with that? I think you can do a whole lot, and smart people will do a lot. But almost by definition we don't know what it is yet because it hasn't been done. But if you think of constraints on IT experimentation within companies, a lot of them have to do with the fact that traditionally there's been a lot of upfront expenses required to build the capacity to do experiments, to write applications that may or may not pay off, and so that cost squelches innovation. And suddenly those costs are going away, so there can be a lot more experimentation going on than we've seen in the past. That's just looking within IT departments' purview.
Beyond that, when you look at Amazon's Kindle the most interesting thing about it isn't that's it's another e-book reader, but that it has a perpetual Internet connection essentially built into the product, by which I mean there's no extra cost, it's just a feature of the product. What happens when more and more products have Internet services essentially bundled into them and you don't even have to think about it? It seems inevitable that's coming, and companies are going to have to think about how that changes the nature of products and services when they're always connected to this incredibly powerful cloud.
Now, this idea that such a small number of companies are buying such a large chunk of servers -- what are the implications?
Well, first of all this comes from a guy at Microsoft [Rick Rashid, senior vice president who oversees Microsoft Research]. So I'm assuming he has good information and that's an accurate number. But if that is, 20 percent is a huge chunk of the server market, and to have that consolidated into the hands of a few purchasers in what's only the last few years, really shows a fundamental shift in the nature of that industry that more and more of the product is going to be consumed by fewer and fewer companies. That radically changes the server industry.
Page BreakIs it surprising to me that that trend is underway? No, it reflects the fact that more and more computing is done in central datacenters now. Forget about the corporate world. If you look at how individuals use their PCs or smartphones today, a huge amount of stuff that used to require buying a hard drive is now done out in the cloud. All Web 2.0 is in the cloud. It doesn't surprise me because it reflects how people view the way their computers work. But if it's already at 20 percent that seems pretty remarkable in such a short span of time.
In what ways are you seeing enterprises use Web 2.0 today?
The value of that model isn't limited to college kids on Facebook trying to find dates. There's a lot of opportunity for companies to take this Web 2.0 model that builds on shared systems, the ability to provide the user with a lot of information, and tools that enable them to control what information they share and who they share it with at any given moment. It really replicates the fundamental aspects of business organizations where a lot of what you do is figure out which colleagues have information that you can use, how to share it with others, how to get information into the right hands.
We really haven't seen powerful social networking tools evolve for individual businesses. It will be a generation shift. People who are so plugged into social networks at home or at school, they're going to want those same capabilities at work. It's really driven by the user because it upsets the traditional IT apple cart. IT departments and staffers will generally drag their feet and then will play catch-up.
Vivek Kundra, the new federal CIO, has said that the personal technologies he uses are so much better than what he was using professionally that he just had to adopt things like YouTube for the D.C. city government...
Quite a while ago, I was interviewing Marc Benioff about the origins of Salesforce.com and he had a very similar story. He was working at Oracle at the time but using things like Amazon.com online. And he said, "This is really powerful and I can do all sorts of customized stuff with it. Why can't I do this with enterprise applications?" In the story he tells, that was the inspiration for Salesforce. You can relate to that because when you compare most corporate applications to what you find a-dime-a-dozen of online, computing is much easier through the services you get online everyday than it is going through your traditional corporate applications.
One last question for you. What is the most significant thing enterprise IT shops should brace themselves for?
The big thing they'll have to brace themselves for is that the functions that until now have accounted for most of their spending and most of their hiring are going to go away, such as all the administrative and maintenance jobs that were required to run complex equipment and applications on-site. This isn't going to happen overnight, but much of that is going to move out to the utility model over time. That doesn't mean IT shops won't continue to exist and have important functions -- they might have even have some more important functions -- but it does mean that their traditional roles are going to change and they're going to have to get used to, I think, having a lot fewer people and probably having considerably lower budgets. Again, I'm talking about change that will play out in 10 years, not change that's going to happen in two years.