If you follow tech news at all, you’ve heard by now that Google has officially stated that they will be launching an operating system targeted for netbooks. There was an incredible amount of speculation and talk about when Google would announce a general purpose OS and it seems like that time has finally arrived. This is big news because unlike more conventional operating systems like Windows, Mac OS or Linux, Google is aiming to make Chrome OS the first consumer “WebOS”.
This stems from the idea that more and more of what we do with computers is done online through a web browser and eventually, one day, everything will run through the browser – including the operating system. We have gotten a glimpse of this trend with great applications such as Google Maps, Gmail, and even Facebook which are all rich web applications in one form or another.
Now that’s the rosy part. At the risk of possibly eating my words in years to come, here’s why there is absolutely no way an on the cloud web- only OS will work:
The internet needs to be fast, ubiquitous and always on which is currently far from the case. Sure, we have access to wireless networks when in range of a cell tower which is a good amount of the time. Then again, there is a fair bit of latency involved with that sort of a setup and tunnels still exist last I checked, as do basements. No signal anyone?
Client systems are getting more and more powerful. So what you may ask? Well many years ago when computers filled a room and cost millions of current-day dollars it was decided that to save money you could build just one machine and have people access it from terminals. That idea still exists today with so-called “thin-clients” which are used to save power and money by running cheap machines off a fast server. That said, as client machines get more and more powerful, the advantages of centralization decreases – not the other way around.
A cloud model is inherently bandwidth-inefficient because you need to transfer the same data from the client to the server over and over again. Think about it: if you put the OS on an online server then you need to download parts of the OS over and over again rather than maintaining a locally stored copy. No matter how you look at it, local data is faster and more bandwidth-efficient. Even in the datacenter, when you want to create a SAN (Storage Area Network) you need to be very careful of latency on even very fast networks.
Working exclusively in the cloud is neither desirable nor practical. As some may point out, many applications such as Facebook require that all information be centralized in the cloud to work, but there are many cases where that need not be the case and is even a bad thing. Take video editing, for instance: it has both huge compute and bandwidth demands. Working with a local copy is just that much easier and more practical. The same holds true for most compute-intensive applications.
Many businesses and individuals will resist storing and working with their data in the cloud because they want to protect their intellectual property. Wouldn’t it be just great if the government dropped Windows in favor of web-centric operating systems and managed all your data with web-based operating systems? Would you really want to manage absolutely all your data on somebody else’s system? Besides, we all know that Google will be sifting through any data that we store on their servers as they already do.
All these problems will always leave the Web OS behind the times because there will always be more powerful applications on a local OS with direct hardware access. And even if you make, say, a photo editing application which draws on local resources, an installed copy of Photoshop will always load faster. For the same performance you need to water-down the web applications; just try out Google Apps once and you’ll see why Microsoft Office is and will continue to be dominant for years to come.
A Web OS does not eliminate the need of some form of hardware/software interface which is a role the OS currently fills. Chrome OS will need a Linux kernel on the clients with drivers managing hardware that the web browser will draw upon. Why go to all the trouble of creating a hardware interface but not including commonly used applications such as, say, a media player, a word processor, an image/video editor or even a hypervisor like we already do?
So to sum this all up, it seems ridiculous to drop locally stored copies of data that we want to keep local and private. It seems even crazier to just remove software which draws upon fast local resources and can be used over and over again without unnecessarily consuming network bandwidth. Don’t get me wrong, the cloud model is very useful and often even required for collaboration, communication and in many other instances but it can only go so far. There are a great deal many things which can be done only in public but there are also times where you just want to be alone.