In the early days of computing, IT delivered computing using huge mainframes, accepting input from users, crunching the numbers and returning results to a very thin client – just a green-screen terminal. PC’s came into the mix, allowing users a local experience with graphical interfaces and even other larger PC’s as “servers”.
And now comes the “cloud”. It seems everything with a network card, or anything that uses a network card, is “cloud” or “cloud enabled”. But what is the “real” definition of the “cloud”? In effect, it’s just outsourced or virtualized IT.
There are really three levels of this definition. The first is “Infrastructure as a Service” or IaaS. This means you just rent a server or storage on someone else’s network. While this is interesting, you still have to figure out scale, management and so on. You’re only relieved of the hardware costs and maintenance. It’s interesting to IT because it buffers them against hardware costs and upgrades, at least directly. In fact, you’re probably doing this yourself already, by using Hyper-V or VMWare in your datacenters.
The next level is “Software as a Service”, or SaaS. You can think of things like Hotmail or Expedia in this way. You log on to an application over the web, use it, and log off. Nothing to install, no upgrades, no management, no maintenance. This is also an interesting “cloud” offering, although the limitations are the same as the mainframe days – you can’t easily make changes to the software, and in some cases, you just get what you get. Great for consumers, not as interesting to IT. In fact, for us IT folks, this is what we need to provide, not necessarily something we use.
Finally there is the “Platform as a Service” or PaaS. In this model, you have a platform to write on, which relieves the hardware issue of IaaS, you can write code that delivers software as in SaaS, and provides a way to create, test and deliver both. You’re insulated against hardware and even platform upgrades, and you’re given a way to scale the application without caring where it really lives. This is the road that Microsoft takes with Windows and SQL Azure.
Of course it isn’t all rainbows and candy – there are things to think about, like security, uptime and more. There’s a formal process for determining the applications that fit – which I’ll write about later.