First, an apology: I am a slave to alliteration. I could also have gone with "Cloud casts its own shadow" or "Clunky cloud crumbles" so really, I think you have to thank me for what little restraint I did manage to exhibit in writing the post title.

Like everyone else, I am referring to the Gmail outage this morning, which had little effect on mainland US users but cast the rest of the world into disarray for a few hours. Considering how many people got their knickers in a twist when Google removed the "Search the Web" option from Gmail (I am still getting comments on that post, a * later) I can't imagine what paroxysms this must have thrown loyal users into.

There's no detail on what caused the problem; it's certainly not the first and won't be the last, and considering the low, low rates, I think a rational evaluator would be hard-pressed to say that you weren't still getting excellent value and performance for your money. If you are paying any money in the first place. Which most of you aren't.

So of greater concern to me this morning is the demise of PaaS provider Coghead, which has been purchased by SAP for its technology, but not its customers. The technology, then, is expected to survive, but the customers are going to be out on the street as of April 30.

This is a more troublesome development, because it represents one of the first PaaS implosions to encompass the great fear of all potential cloud computing customers: "what if my provider goes out of business and takes all my stuff with them?" Other providers are offering, as usual, to take up any cast-off customers, but in this case it isn't as easy as pointing your users to a new website and getting up and running in a few days.

Cogshead is following the general pattern of utility computing provider failures so far in that it is making available all customer data and providing free service and support through the termination date, but it is unique in that the real value of the service was not in the data itself, but in the applications developed by customers on the platform. If Microsoft goes out of business tomorrow (down another 3% this morning, I see) you can still take your C# code and compile and use it wherever you want. Not so with your Cogshead application logic; it's worthless without the platform. The same is true, to a lesser extent, for any SaaS operation... while the data may be the most important thing, without the application framework it was ensconced in it becomes difficult to realize that value again.

A lot of people have been calling for portable standards to prevent this sort of dead-end for customers of failed SaaS/PaaS providers. It sounds like a grand idea but I'm not certain how practical it is. There is a lot of innovation happening in this field and a lot of businesses which think they have a competitive advantage based on their own approach, and it's difficult to imagine them giving up their differences to adopt a common definition language. Nor would such a beast necessarily be of tremendous interest to customers... they want stability more than portability, particularly when getting the portability might affect performance.

My guess is that the solution to this will lie more in consolidation and compartmentalization than in standardization. Service providers will be snapped up by larger, more stable companies, or will learn to structure their offerings to allow more flexible deployment relying on those larger, more stable companies. Imagine if Cogshead's delivery had been offered via packaged EC2 instances, for example. Of course, there is a whole other fear there about relying on a proprietary virtual machine format as well, but Amazon, at least, isn't going to fold up on you in a month... so that's a concern that can safely be relegated to the "future nightmares" category.