These days it seems like just about every vendor is pushing its numerous and varied cloud offerings. It also seems like businesses are holding off on moving to cloud computing with just as much vigor. Everyone has heard all the different cloud sales pitches: Cloud services offer unlimited scalability, they provide global access to computing services, they eliminate the need to buy private infrastructure, they transform IT costs from CapEx to OpEx, they reduce management, they allow IT to be greener, and the list goes on. The funny thing is that despite all the hype, IT just isn’t buying it—at least not yet. And this reluctance to jump headfirst into the cloud is reflected by many of the cloud sales specials that you see popping up. Many vendors, such as Amazon, are offering free low-level services to try to get you started with their offerings. Other vendors, such as Microsoft, are offering free trial periods to entice you to try their services.

Related: SQL Server Cloud Choices

One of the reasons businesses aren’t all gaga about cloud services is that these services harken back to an era of computing that fell out of favor a decade or two ago—the era of the mainframe. If you think about it, cloud computing is a lot like the old centralized IT processing model of the past embodied by the mainframe. Computing power is moved away from the end users and into a centralized entity that’s managed by someone else.

To be fair, the cloud isn’t really centralized the way mainframes of the past were (and are). However, the cloud and mainframes are alike in that they both move the computing power and infrastructure governance farther from the users of those platforms. Given the nature of cloud computing, the mainframe appears to be something like the ultimate cloud platform. Extreme scalability, availability, virtualization, and usage-based metering have been part of the mainframe for decades. And this mainframe mentality is part of the problem many organizations have with the cloud.

Businesses moved away from that model of centralized computing on purpose. Businesses, and perhaps more specifically their departments and end users, wanted more control of their computing resources. They didn’t want to share them with other business units—let alone other businesses. This is exactly what the cloud requires that you do. These factors led to the rise of distributed PC computing, client/server computing, and the web that we use today. Somewhat ironically, Microsoft was one of the leading vendors driving the push behind the distributed PC computing model and away from the centralized computing model, but today Microsoft is one of the leaders in the push to the cloud.

I know that cloud vendors would adamantly disagree with this assessment, but in some ways moving to the cloud is a bit like a step back in time. Cloud computing definitely moves much of the control of the computing resources away from the end users they service. The cloud also moves control of those resources out of the hands of IT. This isn’t the first time an attempt has been made to resurrect this centralized computing model. About a decade ago vendors tried to sell IT on the then-hot trend of thin-client computing. IT didn’t buy into that one, either—although perhaps in the long run it did herald the transition away from desktop applications to browser-based applications.

Perhaps computing trends are just like fashion, and just like those old bell bottoms from the 1960s have come back, so will the centralized computing model—as the cloud but with a modern twist. Certainly the landscape today is vastly different than it was in the era of the mainframe, or even a decade ago when thin-client computing was the media darling. Network connections are better today and the Internet, though far from perfect, is more reliable. But the $64,000 question around the future of cloud computing is, What do businesses really want? Do they really want to move back to a more centralized computing model in which the processing power and the control of IT resources lie outside the walls of the business, or do they prefer the more distributed, albeit more complex, model of distributed computing that we have today?