Published: 28 Oct 2010
As cloud computing takes hold in the enterprise, changes to the normal patterns of what is both art and science in IT are expected to shift dramatically. Given that this business model is so different from today's hierarchical, client-server model, skill requirements will shift radically to suit the demands of a services world.
Applications going out to the cloud smell like outsourcing all over again.
Vasanthan Dasan, CTO and VP of engineering at The Armada Group,
The concept is nothing new. Rare is the person who has to write his own printer driver today. Twenty years ago, there was someone in the IT department who did just that. The same goes for network interfaces and network and routing topology. Even application development has undergone some hardcore simplification and standardization; within living memory, programmers wrote instruction directly to their processor, one line and one instruction at a time.
How much change do we need and when do we need it? Experts say the best approach is to go slow, determine your end user's needs and adjust skills from there.
"The main issue for us is figuring out how [cloud] changes how our people use technology, so we are looking at staff functions and how it will change for them," said Brett Michalak, CIO at Crescent Healthcare, a pharmacy company that provides home healthcare services. Though today's version of cloud computing is relatively new, Michalak said it was often hard to pin real business value on cloud services, creating confusion and skepticism among staff about its future impact on operations.
Michalak said his firm would like to deliver some apps as services to mobile devices for his end users, but he was going to wait and see how other large organizations handled their transitions. Few solid case studies exist, which makes cloud computing concepts hard to sell to the boss, he said.
Randy Bias, CEO of Cloudscaling, helps big firms like Kaiser Permanente build and use cloud technologies, and he said he has seen the process of shifting jobs descriptions up close. He said he believes the days of the traditional application, server and management hierarchy are winding down.
"I like to think that cloud is displacing enterprise computing in the same way that mainframes were displaced," he said. Mainframes didn't go away when the client-server model came in; in fact, mainframes still run some of the world's largest institutions and COBOL still has billions of lines of new code written annually. "What happened is they became a niche," Bias said.
More on IT skills and cloud:
He predicted a similar transformation within the enterprise; some traditional server deployments will remain because they will continue to be the best way to run some applications. Eventually, however, he said he believes the idea of IT consumed as a fungible, automatically generated and consumed service will predominate. Bias said that IT staff will want to become familiar with how the basic cloud model functions and how to interact with it.
Flexibility and low touch but still some server huggers
Bias compared it to the automobile industry, where assembly line workers were replaced by robots and the jobs that went with that were robotic specialists and managers. The hands-on IT work done today will get rarer and there will be a flattening out of IT and project development, where IT departments simultaneously get more flexible and less hands-on.
That's not to say that many aren't worried. Cloud services providers are aggressively pushing their services, and the sector continues to advance. IT jobs are still prime targets for layoffs, and cloud computing may look like an opportunity to whack jobs if you are a bottom-line budget axeman.
Vasanthan Dasan, CTO and VP of engineering at The Armada Group, a Santa Cruz, Calif., firm that helps companies retool for the cloud, says a lot of people are concerned about job security. "We're in a recession, or coming out of a recession depending on who you believe, which means people are still worried about jobs," he said. "Applications going out to the cloud smell like outsourcing all over again."
"The benefits of using cloud computing don't show up until the enterprise can pool its resources into a single, monolithic unit that can operate efficiently," said James Staten, a principal analyst at Forrester Research in Cambridge, Mass. "There would be protracted cultural and political battles with so-called 'server huggers' and IT denizens jealous of control or scarce resources or budget."
Forrester said as few as 5 to 6% of companies are cooperating in this manner, but it comes from an executive edict that says to all departments, "thou shalt get along," Staten said. He said that "virtualization maturity" would push the trend along until IT staff would eventually see that he (or she) who holds the reins of the cloud is the one with real influence.
The main issue for us is figuring out how [cloud] changes how our people use technology.
Brett Michalak, CIO at Crescent Healthcare,
"As companies gain more experience, the IT department realizes the cool job is the one where you manage the entire pool and set policies that automate a lot of the manual work, so that you're not just provisioning virtual machines all day long," Staten said.
Skills and understanding of how these things get automated and virtualized are going to matter. In large part, much of the job shift will be about keeping up with the new products and technologies, a standard balancing act for most IT pros. After all, many of the things an IT shop buys these days were done by hand not long ago: ticketing systems, invoicing, email, operating system and software management, and monitoring tools -- all well within the grasp of your skilled admin who can code a bit. But who wants staff slaving away for days or weeks on an application that can be bought for $500, or $50? It's not worth the time.
As computing power and network bandwidth gradually turn into an automated, pooled and accessible resource, IT pros will come to think of servers the same way they think of their helpdesk system or their VPN endpoints -- pre-built, pre-configured consumables ready to be used and abused, and replaced or discarded at will.