Years ago, core computing tasks were performed on shared mainframe computers, with individual users assigned terminal devices to communicate with the mainframe computer. Some of the terminals were smart, and able to enhance the interaction. Notably, the CTC Datapoint 2200 was in fact itself a programmable computer (in production five years before Steve Wozniak's Apple I), and was the direct ancestor of the x86 computers of to-day. But few of smart terminals themselves ran any code other than to provide interface for communication with the mainframe. (And the dumb terminals ran no applications.) There were efforts to get the general population using mainframes by way of terminals located in their homes, but these efforts enjoyed limited success.

Then the idea of personal computers caught hold, so that a large share of the population indeed computed at home or in the office. But the computing itself was primarily done at basically the same physical location as was the user. It was possible to add some communications hardware to the computer, and then use it as a terminal device, but most of the tasks that had previously been performed on a mainframe were now being performed locally.

When the 'Net came into wider use, some people started having the thought that perhaps it would be an advancement if principal computing tasks were moved onto the Internet, which is to say onto serving computers that were available by way of the 'Net. Unsurprisingly, I see this as a return to an earlier, previously unpopular model.

Now, sometimes, changes in an infrastructure can breathe new life into essentially older technologies; and one shouldn't reject this idea of moving back to locating core computing on remote machines simply because we had previously reduced its relative use. But I find it a signally unappealing idea, because it removes the independence of personal computing. I very much like the fact that I can do everything without communicating except communication itself. I have local applications for text processing and for type-setting, for multimedia generation, for mathematical analysis, and for programming. For these things, I don't have to rely upon a connection to the 'Net nor upon someone's server.

While there are some tasks that might be better performed by a network of distributed service, there is no particular reason for handing responsibility to such a network for mundane tasks that users could easily be performing with local equipment. And the introduction of the opaque buzzword cloud to refer to distributed service on the Internet does nothing but get my back up.

Anyway, I was prompted to ventilate by this story:

T-Mobile Sidekick users have had things such as contact information and photos stored on the cloud, which in this case is to say some servers that T-Mobile has leased from a division of Microsoft, which division is aptly named Microsoft/Danger. Well, the servers weren't properly backed-up, they crashed, and most or all that they held is just … gone.

[Up-Date (2009:10/15): Microsoft has now largely reversed itself, declaring we have recovered most, if not all, customer data for those Sidekick customers whose data was affected by the recent outage. We rebuilt the system component by component, recovering data along the way.]

Tags: cloud computing, Internet, Microsoft, T-Mobile

This entry was posted and is filed under commentary, information technology, news, public. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.