I was recently at a HP conference in Frankfurt, Germany, where I had the opportunity to talk about the so-called “dinosaurs” in IT department, you know, the one who says that “all this cloud stuff has been going on for ages and just has a different name now.”
Naturally, my HP hosts were not inclined to agree: to them, cloud computing represents a revolution in the relationship between business and information technology, and has no real historical parallel. The argument that people consumed IT as a service as far back as the mainframe era is specious, they say, because of the scalability and dynamic provisioning that characterizes what we call “cloud” today. That’s something new.
And these IT folks were in for a rude awakening if they didn’t get with the program, said my hosts.
If you look at the past 60 years, there has been a cycle of sorts, from buying computing as a service to owning your own gear, and now back to computing as a service. The cloud naysayers, I presume, would have us believe that this cycle will continue, with on-premise computing becoming popular once again at some point in the future.
The problem with predicting what the IT landscape will look like in 10 or 20 years is that we don’t really have enough historical experience to base our assumptions on. The mainframe era is so fundamentally different from the Internet age that I don’t think it’s reasonable to draw anything but the most basic parallels between the two.
True, in the 1960s and 1970s IT was delivered as a service, as it is now, but the issue of cost was the key factor. Many companies simply could not afford to buy their own computers for the same reason they couldn’t afford their own private aircraft. Outside of very large enterprises, the cost-benefit ratio favoured a service model. Nobody chose a service model over a on-premise model; they just bought what they could afford.
That said, it’s arguable that we will see a cycle go on. At least in some ways — I can’t, for example, imagine IT departments foregoing virtual hardware for physical hardware down the road. That would just be silly (though in fairness to dinosaurs, they are right that “virtualization” is nothing terribly new).
Nor is dynamic provisioning and scalability of resources ever going to be a bad thing. Those are both features of cloud that are here to stay, I think.
But ultimately, with improvements in technology, I can see the “as a service” aspect of cloud giving way. Here the dinosaurs may be right. I imagine that many enterprises today would prefer to deploy their own private clouds, but realistically, it’s just not cost-effective unless they have the scale. But once the underlying hardware is cheap enough, and systems become much easier to deploy and manage, it may indeed become the dominant model.
And an on-premise private “cloud” is only referred to as such because it resembles a public cloud within defined limits. In a sense, it’s really only analogous — rather than equivalent — to what we know as a public cloud. “Billing” is replaced by analytics and “service” is actually delivered by an internal IT department.
So, while I have no dog in this fight, I suppose I can understand the dinosaurs’ argument: Servers and personal computers shared some of the characteristics of mainframes, but we never called them “private mainframes.” Nor do we still refer to cars as “horseless carriages.”
I don’t think cloud computing will go away. I can see it live on for many more years, perhaps only in private environments, but if we’ve learned anything from our brief IT history, it will eventually be called something else.