Archive for September 15th, 2009

ChromeOS, Jolicloud, and web desktops

There is very little that Google says that is not analyzed to death, and that sometimes leave people puzzled. The announcement of Google Chrome-OS, a Linux-based lean operating system designed to streamline the use of web-based applications (especially for NetBooks) left a few scratching heads in the blogosphere, and was promptly dismissed by Microsoft as irrelevant. It is true that examples of the same concept abound, like the nicely executed Jolicloud or the various Ubuntu netbook remixes; at the same time, the clout and market power of Google has of course an undeniable impact.

The interesting point is that more than the idea of a lean Linux desktop, the fact that an enhanced web browser (along with some additions like Flash, HTML5, Gears and whatever) can nowadays be considered an effective desktop replacement is something that just one or two years ago could have been considered heretic, and with good reasons. But now, I do most of my writing in Zoho (that is in terms of features much better than Google docs), I use my Zimbra web-based client, I have a java-applet for SSH and even a Quake Live account for the occasional fragfest.You can do video editing, play music, watch TV, code, and there is no doubt that the amount of things possible within a browser will only increase.

The interesting point is that having a web-based infrastructure provides an alternative to all-out virtualization, thanks to the almost stateless approach that is typical of HTTP; having most of the processing handled by the browser reduces the server-side costs of providing services of one order of magnitude or more, while facilitating things like high-availability and in general accessibility. Not only that, but application provisioning become something simple and comprehensible, easily enhanced by the various strong single-sign-on system that are now available (and open source, like OpenSSO).

The browser, along with the innumerable additions that are now used, has become a good enough platform for computing for the mythical 95% of the population-and the cost savings of using a transaction-based architecture when compared to desktop-based (and pixel-based) rendering makes it very clear that the approach will continue to be explored.

The announcement (and, I hope, near future release of ChromeOS) will not in itself mark a significant change in the landscape, at least not without a substantial support (for example, as part of the BIOS of netbooks) of hardware vendors and an increase in availability of cheap and unmetered (or nearly-unmetered) bandwidth. It may, however, create a co-marketing opportunity that can be leveraged by mobile and converged telcos, for a remotely-managed, secure and extremely cheap design. Such a design can be extremely effective for business users, that need security, manageability and independence – all through a standard web browser. And if traditional pixel-based remotization is still necessary for legacy applications, it is still possible to export them, safely tunneled in an HTTPS connection, through open protocols like SPICE or RDP (eventually compiling the viewer as a native client application, so it can be delivered safely along with the connection).

Even if ChromeOS is not successful, I believe that within 2 years the concept in itself will be so economically compelling that it will make desktop virtualization marginal at best.


1 Comment