Dvorak makes it seem binary. You either have a local solution or you have to rely on a web server. I bet many successful apps in the future will have a slightly thicker client (e.g., based on Adobe AIR, Microsoft Silverlight, etc) that syncs with the cloud but still operates locally. Google Gears is one step in that direction.
There's a (I don't know whether to call it a fallacy, but I will)...there's a fallacy that Jumbo Jets are safer than automobiles. This assertion is easily derived from looking at death statistics from each vehicle. But then, such statistics involve random parties. You or I aren't random drivers and so we may be very safe or very dangerous on the road [1].
Anyway, the average person's data is much more likely to disappear than that on google's server clusters. But if you are diligent, you can have backups that work 99.9999999 percent of the time quite easily.
[1] Of course, on roads, the guy driving in the other lane is pseudo-random, so... But there is no other lane with computers.
I am ALWAYS concerned about this (some may call it fanatic). I back up everything twice every night. Hit the safe deposit box once a week. Always have one thumb drive with me, another hidden. Print hard copies of recently changed code.
What if I have a break-in? What about a fire? How about lightning? (Already lost 2 servers that were protected - it happens.) What about a flood?
Sometimes I wonder why I don't just store everything on the "cloud". One of these days...
Source code shouldn't be a problem. It practically backs itself up: 1) Production server(s), 2) Test server(s), 3) Development server(s), 4) Source control server.
A bit to many 9's. 99.9999999 uptime means a downtime of less than a second per year. If my calculations are at least close it's about 1 second downtime during a period of approximately 30 years.
As far as cars go, I remember reading something about 80% of the people thinking they are above average drivers.
Not talking uptime/downtime but just the ability to restore lost data. If I have 3 backups, what are the odds that all three will fail at the same time as the source (assuming different locations, which is the case with my own system.)
A big problem with the article is that they throw up "Trust" as this gigantic, vague moral litmus test and anything you "can't trust" is evil and To Be Avoided. In typical fox news fashion the article is so laden with presuppositions that it's hard to discuss rationally.
There's an important distinction between what happened with WGA and the "Software as Service" model in general. Namely, that WGA is a mostly superfluous add-on only there to protect against piracy. The fact that the system shuts down or locks up when WGA is not available does nothing to support what the users bought the OS for in the first place.
It's true that you can't trust 100% uptime from web servers. Millions of Myspace users are pretty solid evidence that people can live with this. The advantages of the server-based method are generally appreciated by customers, who are also generally aware of the drawbacks. In contrast, users tend not to appreciate being treated like criminals and having their OS crashed on purpose because it couldn't complete a minor validation step.
The crazy premise in this story is that MS represents great technology> MS radically messed up their online service > therefore online on-demand services are a bad idea. Yikes!
i think this article takes what happened to WGA in the wrong direction. I think the lessons to be learned from the WGA outage is that critical systems should avoid relying on external services or at least provide a fail safe mechanism.
Also, as far as I can tell from the article the servers that went down were not hosting true web applications that users log into, just web services that some desktop software relied upon. There are plenty of other network services that desktop computer need to function correctly (for example DNS). Perhaps the WGA failure should server as a lesson in risk management and designing applications with decent failure modes.
Which is more likely to happen?
a. Google, Amazon, Yahoo, or <other host> goes down and loses all of your data.
b. Your hard disk crashes and you have no backup.