Cloud computing is all the rage in the IT industry these days, but there's still plenty of confusion about how cloud applications work and what kind of long-term impact they will have on business technology.
In this article we'll take a look at six common misconceptions IT pros and users have about cloud apps, related to performance, security, management, cost and the long-term effect on IT staffing.
1. Cloud computing will put IT pros out of a job
It's true that shifting internal IT functions to cloud applications lessens the need for a business to have a huge IT staff. Some organisations have moved to cloud services such as Google Apps because they no longer had the staff to handle an internal email and collaboration system. This cause-and-effect may be switched in other cases, with layoffs occurring because of a move to outsourcing.
But the long-term risk to IT employees may not be as terrible as it might first appear, analysts say. For a business that uses many cloud apps, internal IT staff is still needed to manage and integrate the services. Some IT managers may start to see themselves more as vendor managers.
"There's no question it could reduce the demand for traditional IT skills," says analyst Jeff Kaplan, managing director of Thinkstrategies. "But that doesn't mean there can't be a whole new generation of requirements around vendor management, to properly evaluate, select, monitor, manage and contract with these cloud computing companies."
These "traditional" IT pros might find their roles shifting to business unit support, making sure every employee has the right cloud tools to do his or her job. Another obvious point is that, over time, as technology resources shift from customers to vendors, the vendors will have a greater need for IT pros, creating more jobs on the vendor side.
If you're an IT administrator managing servers, applications and other in-house systems today, a shift to cloud computing could force you to learn new skills or eventually find a job elsewhere. But smart tech people already know their industry undergoes constant change, and should be ready to adapt.
2. The cloud is free (or at least incredibly inexpensive)
The Google Apps business version is $50 per user per year. Microsoft's online services start at $120 per year, and both Google and Microsoft offer free versions to home users.
That sounds inexpensive, because it is, but users and analysts are quick to point out that licensing is not the only cost of using a cloud service.
"One of the biggest preconceived notions you have is that it's free. It's out there, it's not a big deal, just sign up and everything's great," says Scott Weidig, who is a technology coordinator at Schaumberg High School in Illinois and uses a variety of cloud apps for business and personal use, including Zoho and Google Apps.
Some customers report having to upgrade Internet bandwidth to take advantage of cloud services (more on that later in this article).
There are also hidden legal costs, because there is generally an increased risk profile when you move applications from the enterprise data center to the cloud, says Nolan Goldberg, a patent and trade secret litigation attorney for Proskauer Rose LLP in New York.
"Cloud services will not necessarily be cheaper," Goldberg says. "The misconception is that a cloud service is always going to be cheaper because people are looking at the sticker price, and not the true total cost."
Before adopting a cloud service, a customer should perform due diligence to make sure the service is suitable for a given application, and that the contract with a cloud provider is not harmful to the customer. Some cloud service contracts state that the terms can be changed by the provider at will or with little or no notice, a troubling development, Goldberg says.
Storing data in the cloud brings new security and litigation risks, but unfortunately it is difficult to calculate beforehand how those risks will increase the customer's potential financial obligation.
3. Cloud performance is never a problem
Just because you use a cloud provider doesn't mean you can ignore your own IT infrastructure and expect great performance. Information systems director Timothy Porter says after adopting some cloud apps a few years ago, "we had to reevaluate the amount of bandwidth we had in our corporate office."
Porter is the IS director at WS Development Associates in Chestnut Hill, Mass., a real estate developer and owner with 82 properties throughout New England.
WS Development Associates users cloud software applications for invoice processing, payroll and other HR functions. The company originally had two T1 lines, and upgraded to a Verizon Mux setup for data rates of up to 50 megabits per second. Cisco WAN acceleration devices were also necessary to improve performance at branch offices, Porter said.
Before the upgrades, performance was a major problem.
"When you start processing a lot of invoices via an application that is not hosted internally, you will have a lag while doing the processing," Porter says. "When you're trying to process dozens of invoices, that really adds up over time, and user frustration can show itself."
Even if your Internet bandwidth is solid, other factors can degrade performance. At Schaumberg High School, Weidig says a Web filter installed to prevent students from surfing inappropriate sites was slowing legitimate educational websites, including a cloud-based reading assessment tool used by students. It turned out the hosted reading site was using nine different IP addresses and URLs, and the Web filter only knew about two of them. Reconfiguring the server to more efficiently allow access to trusted sites solved the problem, but identifying the issue was not a simple task.
Cloud vendors like to make it seem as if cloud platforms offer infinite scalability, but that may not be the case for power users. Spirent, a network performance testing firm, recently evaluated several cloud providers including Amazon, Blue Lock, Hosting.Com, GoGrid, Maximum ASP and Rackspace. The company found that as the number of connections to a cloud service goes up, response times also go up. This is perhaps not surprising, but some of the response times lagged to as much as three to five seconds, says Tim Jefferson, general manager, networks and applications group at Spirent.
In order to serve an unlimited number of users, "what you have to give up is performance," Jefferson says. "As you scale up in one vector, it may cost you response in another vector."