A few years ago we were talking about grid computing and Web 2.0 services. Today we talk about Cloud Computing as an evolution of those terms applied to the new technologies usually based on working through a Web browser and the Internet.

New books, newspapers, and magazines pushing us to the Cloud and its applications appear every day. However, there's a privacy concern in sending our documents to someone else's.

This concern was raised - among others - by the Free Software Foundation and its founder, Richard Stallman: If we send our private documents, music, pictures, and videos to big company servers, we can lose control over them. And what's worse, we're giving control over our digital lives to what will probably be the next giant monopoly in technology.

If we're a home user, this probably won't be a problem. We're already on social networks and Web services that reserve some rights over our information, but for professional people and companies the situation can fraught with dangers. Not just because we'd be putting our work outside the company, but because we'll depend on those servers to work and on our DSL connection to access that information. And we're not even talking about sending our customer's information there, which could be a legal issue in some countries.

Looking for possible resolution, one way to bring Cloud Computing under some kind of individual control and ownership is through smaller in-house clouds that are linked together instead of using a big Cloud Computing provider where control of data is dicey.

It's difficult to find a unanimously agreed-on definition of Cloud Computing, but one is that the Cloud is the new way of working that changes the schema of everything that's localized on every computer (the OS, Office Suite, media players, files...) to something centralized, where data and files are in a server and users just load them through the browser.

The question is: should we centralize the data of lots of companies in a single mainframe or in a server per company? Microsoft, Amazon, or Google would probably choose the first option, but looking after our privacy, we might prefer the second. That is, one per business, one per network, but small networks, not the Internet, especially for professionals - preserving some control.

But are hosted solutions valid Cloud Computing systems? Do Cloud Computing definitions talk about size? We can easily convert a local server inside a company or home into a small but capable Cloud Computing provider.

For example, in a family of five people the son can set up a server with two small hard drives and the ZFS filesystem (which can be easily expanded) and install software that will let the users work from anywhere connected to that server through a Web browser. In this environment, the 16-year-old son will become the Cloud provider. His mother, father, and two brothers will become the Cloud's users using that service from work or school. And the data won't leave the house.

Whether this is a valid Cloud solution or not, we should certainly accept it as one. Because if we don't include small Clouds as part of Cloud Computing, the Cloud will degenerate into a storm in which five companies fight to control thousands of companies' data. And this couldn't be good. To host our applications and make them accessible from anywhere in the world, usually from a Web browser (one of the most accepted Cloud traits), would only take an expandable server - say with a ZFS filesystem - and a piece of software that would let us to work from the Cloud - this time a Cloud inside our infrastructure.

Just as Linux appeared to offer an alternative to proprietary operating systems, the eyeOS project, a free open source project started in 2005 creates a Cloud operating system that can be easily hosted from almost any Web server. It currently has nine active communities in nine countries. Once installed on a server it provides a Web desktop for each user with some base applications to work with office files (just as we do with Google Docs & Spreadsheets), PIM, file manager, etc. And we can always develop new applications or use the community-provided ones, available through free repositories. It's free software (licensed under the AGPL v3 license) and easy to set up. With that, any old computer that can be used as a server to be our Cloud provider.

On the other hand, there'll be some gaps in this small cloud model: backups will be more difficult to control, and the scheme will probably be a bit more inefficient than hosting everything in a giant place that services a thousand companies. But every one of them will continue working the day that DSL stops. And they will still be working the day that one of the Cloud providers closes its services because of bankruptcy. And the thousand companies who saw the big Cloud provider as the solution to their needs will be trying to retrieve their files from old local copies.

Resources

• http://linux.sys-con.com/node/587717
• http://en.wikipedia.org/wiki/Cloud Computing
• http://www.eyeos.org

• Cloud BI Guru Believes Great Companies Can Be Built in the Cloud