An interesting article from GridToday reporting on a talk by Nicholas Carr, included below in full.
Utility computing has sparked imaginations. Eliminating capital investment, provisioning resources dynamically and paying as you go for what you actually use are very attractive features. Recovering automatically from hardware failures alone could save many enterprises millions. Many considered this Utopian concept unobtainable, but academics have operated this way for more than a decade through the use of grid computing. Unfortunately, the benefits of grid computing were limited to technical computations and required code written specifically to run on the grid. Mainstream business applications have been locked out. Until now.
Recent advances in grid technology have enabled the first true utility computing services. For the first time, companies can build, deploy and scale complex online systems without owning or operating a single server. More importantly, this simplification allows enterprises to regain control of their IT platforms, enabling them to accelerate business-critical updates, features and additions in support of revenue objectives. SaaS and Web 2.0 companies, along with forward-looking enterprises, already are starting to take advantage of this new business model. Gone are the time, cost and hardware barriers associated with bringing transactional Web applications to market and scaling them as the service demand grows.
In an Xtalks Web conference on April 11, Nicholas Carr, former executive editor of the Harvard Business Review and acclaimed business writer, discussed the emergence of grid computing in a broader, historical context. Analyzing the rise of what Carr called the "Third Age" of IT from a strategic and economic standpoint, he took his audience through an IT timeline aiming to explain the rise and ultimate need of grid computing by comparing the modern IT sector to the water wheel of industrialization.
At the very beginning of the Industrial Revolution, manufacturers had to rely on in-house power sources to supply themselves with the vast quantities of power needed to run their factories (e.g., wind mills, water wheels, etc.). This model was inherently inefficient, forcing companies to spend a good portion of their budgets on labor and maintenance to keep their power sources humming. Carr pointed out that today's IT model suffers, more or less, from the same symptoms. Manufacturers found a solution in the utility model of electricity. All of a sudden, they were able to plug into a grid and access the power they needed on-demand and at a much lower cost.
This allusion led the introduction of Carr's first principle of any business model: "The supply of any business resource will gravitate toward its most economically efficient model." The question he then put to his audience was whether or not current business models were the most efficient for IT supply. Obviously, he believes they are not.
In what Carr labeled the "Second Age" of IT, the current age, businesses rely on what compares easily to what early manufacturers had to rely on for their in-house power sources: the water wheel compared to client-servers. The PC, while extremely personable, also is very inefficient by nature. Businesses have to rely on the same hardware and software as their competitors and hire the same kind of people to deal with these apparatuses, resulting in huge amounts of redundancy. What's worse, Carr explained, is that it's all amplified by this tight connection between hardware and software, the physical and logical aspects of IT.
All of this leads to vast diseconomies in software and labor, and there arises a failure to capitalize on software's economies of size. Carr asserts that, currently, 60 percent of IT labor consists of routine maintenance and support, which ultimately leads to a massive drain on management time. Just like manufacturers at the dawn of the Industrial Revolution, companies now have to focus on their products/services, plus run an internal IT department. He also points out that the price of private IT infrastructure comes out to 50 percent of most companies' total equipment investment. In short, he believes that the current model of client-server business computing is hamstringing IT.
Is there a better way? Carr believes that the client-server model will be replaced, and already is, with a utility model or grid model, just as the water wheel was replaced with a utility electricity model, a centralized shared infrastructure.
Looking at IT from a purely economic standpoint, Carr believes we can begin to see why the utility computing model is so compelling. Carr explains that IT is a general-purpose technology, not a tool itself, but a platform for them. It has vast potential for economies of scale if supply can be consolidated and centralized. Consolidation, however, requires new technologies and fresh business. Fortunately, utility technologies are maturing, according to Carr. For the utility computing model to be successful, there needs to be a "grid" of technologies in place -- high-speed Internet networks, data-processing dynamos like Google's new datacenters, virtualized multi-tenant infrastructure, IT automation so software can replace redundant labor, and sensible metering and pricing. As a result, virtualization will allow IT to break the lock between hardware and software. A "best of both worlds" scenario will unfold through the high efficiency of utility grids and the personalization of the client-server model.
This theoretical IT utopia, although slowly becoming reality, requires precise managing of the transition, Carr pointed out. There are three main tensions to be aware of moving forward: dedicated software versus utility software; internal utility versus external; and how to determine if your company should be a leader or a follower as this evolution unfolds. What should be outsourced and what should be created internally? Depending on the wisdom management displays, although there will be many challenges along the way, Carr asserted there will be huge opportunities and benefits.
If you would like to listen to Carr's Xtalks presentation in full, please visit http://xtalks.com/gridcomputing0704.ashx.
Comments