Holy cow, did you read about this company "The Linkup" losing 45% of its customers' data?! How about they change their name to "The @$%! Up".
First off, let me say that these guys didn't have to be retarded to lose this much data. In fact, there are (were?) probably a lot of really talented people who designed and built this system to avoid such things.
I'm an optimist, so I have to believe somebody raised their voice at a meeting when data was shipped off to some loosely linked company from some past relationship. The finger pointing going on now is exactly what nobody ever wants to see happen to something they built.
Nirvanix says it has not deleted any customer data, and promises that its Storage Delivery Network is immune to the problem that plagued The Linkup. At The Linkup, a "system administrator ran a script that misidentified active account data and disassociated physical files from their owners," Nirvanix says. "This led to files being marked offline in the old Streamload/MediaMax file system when they shouldn't have been." Iverson, meanwhile, claims it was a Nirvanix engineer who caused the data loss.
Hiring managers beware.. if you receive a resume of a Systems Administrator from the San Diego area with a couple of years missing and immediate availability, you might want to ask him if he knows what the -f argument to rm does.
We've all been there at the meeting that produces this type of situation though. Somebody finds some amazing 3rd party vendor that will do for the company what now takes a little resources, but will, when the company gets successful, take a lot of resources. Rather than scale our business up, why not just jump onto the coat tails of some other already successful business and rake in the dough, not needing any real technology, just gluing things together.
This is the essence of things like Amazon S3 and EC2, and Google's AppEngine. These companies have built massive clusters to deal with their biggest days.. and somebody over there got smart and said "hey, 1% of our architecture is 200 times the power of most startups. We should sell it to them."
The problem is, one day Amazon will need that power, and they won't think twice about cutting off their tiny little source of revenue to make sure they can sell copies of Oprah's latest selection. Their core business is selling stuff over the web, so why should they care if EC2 and S3 go slow, stop working, or function better than most people's multi-thousand-dollar-a-month contracts with ISP's.
So, if something is key to your success, even if its not "your core competency", I say build it into your business model to gain that competency. I wouldn't pay somebody to pack my parachute if I were a sky diver. I'd pay somebody to teach me, maybe even to check it for me, but I want to know that I packed it right when I pull the cord.. I don't want a little flag to fly out saying "Sorry, We gave up our parachute packing business 2 days ago. Might want to start praying.."