Skype is one more example of why to trust the cloud but wear a parachute

Clouds can crash just as hard as data centers; with no backup, you're stuck waiting.

One more in a list of cloud-based services that are a wicked bargain for people using them until they disappear unexpectedly, leaving all those hopeful users stuck.

Servers at the free VoIP-through-PC service Skype didn't crash, exactly, but Windows users were suddenly cut off yesterday morning east-coast time.

The problem was stil the same this morning, though reinstalling or updating to the latest version should fix it, as should an updated version of the Mac version that is due tomorrow, according to the latest update from Skype's fixit blog.

A blog yesterday warned that some Windows users were having problems logging in. It offered a long, kludgy workaround to clean out problematic session data in the shared.xml file, which stores Skype configuration data.

Customers quoted by the British tech mag Inquirer blamed the acquisition of Skype by Microsoft for $8.5 billion, which the two announced earlier this month.

Until the acquisition is approved, the two are operating independently, Microsoft watcher Mary Jo Foley noted.

That doesn't mean there hasn't been some advance integration work going on that might have messed things up.

The problem, which has come up during outages at Amazon's EC2, Microsoft Azure and any number of other cloud-based services, and the apps built on top of them, is that when they go down, customers have no recourse.

With most IT services it's usual not only to investigate the provider's infrastructure to be sure it's stable and redundant and reliable, but to install an independent redundant service in case the primary goes down.

Skype and Sony's Playstation and Entertainment networks get a lot more attention than other clouds when they go down, because of the number of users and the passion with which people approach gaming or gossip compared to ERP, but the problem is the same.

Relying on an outside service – even one that embodies a magical new age of no-infrastructure, no-cost, all-free-lunch computing is a bad idea if you do it without assuming that some disaster of some kind will bring that service down some time.

It could be a corruption in the client-side files, as it appears to be in Skype's case, or a problem in communication between database servers and app servers, as it was at Amazon.

Or it could be a meteor strike, flood, invasion of aliens hungry for the flesh of data-center managers.

It doesn't matter (except to the data center managers and the aliens who find out an all-pizza diet among their feed stock doesn't make for the most savory result on their own plates).

Hire as many cloud or SAAS services as you want. They're a great way to get services without paying to build them.

But build a backup anyway, and make sure it's not another cloud service built on the same infrastructure.

A lot of cloud services are hosted or co-located at the same massive data centers, which provide services to other service providers rather than directly to end users.

If your primary SAAS provider and your backup are housed in the same data center when the meteor comes down and all the sysadmins are eaten, you're still out of luck.

So are they. But that doesn't help you much.

ITWorld DealPost: The best in tech deals and discounts.
Shop Tech Products at Amazon