Archive for October 18, 2011

Cloud Computing will Fail without the Ability to Backup Data Locally

October 18, 2011 11 comments

The last two years have seen an explosion growth of yet another tech buzzword- “cloud computing“. While many naive readers might think that the “cloud” is a new idea, it is in fact the original form of multi-user computing.

You see.. in the era before I and many of you were born, computers were multi-million dollars and multi-ton machines used by a few brave and nerdy souls. The high cost of those machines made individual ownership, even by many corporations, a bit dear. Consequently many of the mainframe systems were shared by users who accused it through dumb or very basic terminals. There is a reason that operating systems based on Unix are light years ahead of Windows and Macs when it comes to handling multiple users and their processes.

While the dumb terminal connected to a very capable multi-user machine has certain advantages (power, capacity etc) it has a major flaw which was obvious to people even in that era. There is always the possibility of one mistake or misstep taking out every user’s data and access to computing. You can increase system up-time and alleviate the risk of catastrophic failure through clever, but expensive, engineering (IBM Z Series) and redundancies. However the fixes are expensive and you are still subject to the whims of the power-hungry scumbag who owns the system.

The PC “revolution” solved this by creating reasonably cheap personal computers with decent data storage and backup ability. I personally suspect that the very fact that you owned your PC and its accessories was a major factor behind the success of the PC over cheaper dumb terminals.

Fast forward to an era that began with Gmail and ubiquitous high-speed internet, when it once again become much cheaper and worthwhile to store your data and perform computations in a large cluster of remote computers. Today services from Flickr, Tumblr, DropBox, FaceBook, Twitter, YouTube to various Amazon “Clouds” are an integral part of our life- yet their old vulnerability never went away.

If anything those old vulnerabilities have been amplified by the current economic and legal environment. You see.. a cloud built of IBM Z class servers with multi-site backup would be pretty close to indestructible- but also really expensive. Most clouds being built today are increasingly made up of customized servers designed and run with an eye on cost-cutting. Cue Taleb’s Black Swan.. Also companies can deny and throw you of their “clouds” without any real possibility of legal redress

This fragility comes at precisely the same time that “cloud” based services from Gmail to FaceBook and DropBox are becoming an important part of people’s lives. They are now closer to utilities than luxuries and indulgences. However even this would not be a major problem if people could easily mirror the data and applications in these “clouds”. You can, for example, easily backup all of your Gmail data on your local hard-drives (main and backup) at pretty much every login, if you choose to.

But what about services which you cannot mirror, host easily or which have DRM? What happens when the “cloud” dies or you lose access to it?

It is likely that we will experience a series of large outages or serious compromise of some important “cloud” based service in the near future. I am making this prediction based not on my deep understanding of technology but of human stupidity, ego and legalistic bullshit. Only after most people experience the joys of trusting “cloud” computing that we might start to develop a better way to storing and accessing data- one in which both your personal computers and the “clouds” effortlessly mirror each others data and functionality as required.