Down To Business: Will Cloud Computing Rain On IT's Parade?
Apps, storage, and other business technology resources are indeed moving from desktops and company data centers to third-party providers. But don't expect to buy everything from Humongous IT & Telecom anytime soon.
The latest and greatest business technology movement is almost always portrayed as inexorable and inevitable, whether it's on-demand computing or information life-cycle management or social networking. So it is with the very latest über-concept, "cloud computing," which holds that applications, storage, and other IT resources will shift from user desktops and company data centers to massive, often scattered server farms operated by third-party providers, accessible by users over the Internet.
In his book, The Big Switch, Nicholas "Does IT Matter?" Carr trots out the utility metaphor. Business technology organizations, he argues, will go the way of the electric power industry of the early 20th century, whereby lots of self-sufficient outfits made way for relatively few service providers. In the case of IT, the likes of Google and Amazon will supply much of the infrastructure, capacity, and services users consume--or so goes this line of thinking.
We've seen versions of this theory from different sources over the years. There was Sun's early notion that "the network is the computer" and its more recent "red shift" thesis on the future of high-performance computing. Oracle, Citrix, and others have long championed a thin-client, server-based model. And it's nothing new for big institutions to sell raw processing power and bandwidth on their computing grids. None of those movements ever materialized on anywhere near the scale their proponents predicted.
But even if cloud computing is as esoteric as the name suggests, it makes intuitive sense. In its most successful iteration--software as a service--providers such as Salesforce.com are delivering from their data centers Web-based application services that are easier to deploy and upgrade than conventional apps. More traditional IT vendors are starting to embrace the service model. Storage leader EMC, for example, last month introduced a platform called Fortress over which it will deliver a backup service for PCs, laptops, and remote Windows servers and ultimately services that map to its expanding applications portfolio. Vendors are also providing system management, business intelligence, ERP, vulnerability assessment, and other IT resources as centrally managed services.
But Web software, storage, and security services aren't synonymous with cloud computing. They're maybe one manifestation of it. Cloud computing is more about spreading heavy data-processing tasks across thousands of interconnected, often commodity PCs and servers, supplying supercomputing power for applications such as search engine queries (thus Google's central role), complex financial modeling, and drug simulations. Trouble is, delivering specialized IT resources as a utility service is a tad more complicated than piping a pure commodity like electricity.
For one thing, as my colleague Art Wittmann, a former associate director of the University of Wisconsin's Computer-Aided Engineering Center, points out, such complex custom tasks require a deeper understanding of the job submission and retrieval system--a huge problem, he says, particularly for companies with large data sets. Wittmann says storage scheduling within the cloud is very tricky, as providers must move terabytes of highly sensitive data very close to the computing resource allocated to the task just at the moment it's needed, store potentially terabytes of equally sensitive results, and then get it all back to the customer securely and quickly and be ready for the next user. That takes very smart scheduling software. "If you get allocated part of the cloud in Chicago and part in Madrid," Wittmann says, "what do you do with the data set?"
There are many more technical obstacles to widespread commercial use of cloud computing, and no doubt many innovative solutions to those obstacles are still to be invented. But add to those obstacles the valid security and privacy concerns that go with ceding control of massive amounts of sensitive data to a handful of third parties, and it's clear that the shift to Humongous IT & Telecom won't happen overnight.
Rob Preston,
VP and Editor in Chief
[email protected]
To find out more about Rob Preston, please visit his page.
About the Author
You May Also Like