Not THE Cloud, But Clouds

There is always the myth, when a technology is new, that the market for it will be limited.

We can all quote from memory Thomas Watson Sr.'s line about there only being a market for “five or six computers,” when his son was betting IBM's future on their being more, or the late Ken Olsen's belief that no one would want “a PC in their home.”

The same sort of thing happened with the launch of Amazon's EC2 and “cloud computing” a few years ago. The capabilities were so huge we figured there would be only one cloud. Maybe two, since Google was already cloud-like. Then came Microsoft's Azure, and pretty soon every Web host and enterprise CIO realized this was the next big architecture. Now it's going into the “enterprise” space.

Clouds let you scale-up an application for use by millions quickly, or grab a huge hunk of computing time for a big job. The combine elements of parallelism, distributed computing and virtualization. They're just a better way of doing things.

So the race is now on to define cloud architectures, in the belief that, as IBM dominated mainframes, DEC minis, and Microsoft PCs, someone will capture the flag here and make a trillion dollars.

Unfortunately for the empire builders open source has gotten in the way. With shared infrastructure lots of people can compete. Everyone now expects it. No one person can capture the flag.

Victory will go to those who can make cloud deployment easier, who can combine resources so ideas can hit the cloud faster. That's where VMWare is going with its Cloud Foundry announcement yesterday .

Cloud Foundry supports a bunch of open source technologies, resulting in what's called a “Platform as a Service” (PaaS). VMWare wants you to think of it as a “Windows for the cloud” (they even have a PC version) that lets you create, test and deploy applications quickly. (And Cloud Foundry will host them too.)

PaaS is just one of many cloud acronyms you need to know if you're to understand the business.

Amazon's EC2 is basic infrastructure, so it's called Infrastructure as a Service (IaaS). (Software like Eycalyptus and OpenStack is designed to open up that infrastructure to easier use.) If you're creating a software application that will run on the cloud, the result is Software as a Service (SaaS). Cloud mavens will go on-and-on about ee-ass, pass and sass but if you remember the meanings (it's like the way military guys use acronyms to create words like Centcom) you can understand it pretty well.

The idea is to take clouds out of the current “hosted” era, dominated by Amazon, Google and Microsoft's Azure, and allow enterprises to build their own clouds in their own way.  That makes sense, because it's how every other platform has evolved in the past. Azure has gained traction as a haven for developers using Microsoft's Visual Studio. VMWare wants Cloud Foundry to be a Java hub.

The one confusing aspect of all this is the name. You see, VMWare already had something called Cloud Foundry, designed to let its Spring developers execute on the Amazon cloud. This one is totally different.  They're even calling that old version “Classic” Cloud Foundry. Just don't call this “New” Cloud Foundry. Call it Diet Cloud Foundry instead. (Atlanta joke.)

What's it all mean to you? Just that regular companies, like Coca-Cola, can now create cloud projects run on their own infrastructure. Corporate applications will be coming at you from the cloud, from big companies, applications those big companies will be able to control and manage centrally. The companies will be spending a lot of money, outfits like VMWare will be making a lot of money, but whether anything really useful or cool comes out of it remains to be seen.

My guess is that in a few years we'll take the idea of sites that scale infinitely, or huge problems that solve instantly, completely for granted.