The last few days I've been trying out EyeOS, a "Cloud Computing Operating System" that is open source, free to download and easy (for the most part) to install. It's really quite amazing, but is it useful?
The geek in me first reacted with astonishment and excitement. And truly, EyeOS is very astonishing in what it accomplishes, and very exciting on a technical level! The ability to very quickly display what feels like a virtual desktop over the Internet is pretty amazing.
Installation is a breeze - literally just uncompress and dump it in your web directory. No database is required; all user settings are stored in XML files and it has it's own virtual filesystem. The only catch is that trying to get it to recognize OpenOffice and Microsoft Office files is kind of tricky. You must install OpenOffice on the server, and even then I never could get it to open a Microsoft Office file, just OpenOffice and it's own native format.
The more I used EyeOS, the more I realized I'd never use it. It's support for spreadsheets and text documents is marginal at best, and to me that seemed to be the best feature. There is a considerable amount of EyeOS applications written that you can install into your own server for anything from Twitter to watching movies, but as a software developer I had reservations as well.
One of the driving concepts behind "Cloud Computing" is the ability to deploy an extremely scalable application for many remote users over the Internet. In order for an EyeOS installation to be scalable of any sort you would need file system replication between multiple backend servers and some sort of failover load balancing proxy on the frontend like Pound. Definitely doable without a huge effort, but the scalability of such an install would be questioned, simply because of the need to replicate it's entire VFS across multiple machines to handle the load balancing.
In addition, your applications would be tied to their "OS" platform, and your users would need to learn a new desktop environment. I'm embarrassed to report that I wasted almost an hour trying to figure out how to select multiple files using it's file manager window in order to drag and drop them into another folder - only to finally give up.
With my team developing large scale highly available applications with PHP I can honestly say I'd rather not have my developers writing applications for EyeOS. The same applications could simply be written as a web service without the EyeOS overhead and requirements. I would also trust scalability of my applications directly to a cluster of LAMP servers more than I would the EyeOS system. I've not tried to cluster an EyeOS install, but I'd be concerned that the VFS system they use might become corrupt if you're trying to load balance across multiple servers.
If EyeOS truly handled Word and Excel documents as well as OpenOffice does I'd be much more likely to consider it - then my users could sync their documents to the server and have very usable access from anywhere, and it might be worth replacing a few custom desktop apps we use with something that runs in the EyeOS environment. Until then, I think I'll steer clear.
EyeOS is definitely the valuable possession that just costs entirely too much to own. I'm certainly going to monitor their continued progress over the coming year.