Geva Perry is starting a series called “The Purpose-Driven Cloud” where it appears that he’s trying to address the aforementioned question by evaluating a number of different attributes that are all technology-centric.  It looks like it should be a worthwhile discussion, although it’s mostly written from a developer’s point of view.

But I think his angle is missing some important elements (although in fairness they may get addressed along the way) that are more customer-driven and business-driven:

• Suitability to task: What is the kind of application that you have?  Is it primarily workflow and transaction oriented like an ERP application or are you doing heavy number crunching and using complex algrorithms like a pricing credit default swaps?  Some cloud platforms like Force.com are great for the former, but wouldn’t be good for the latter. And if you are accessing data frequently, cloud storage options like Amazon S3, might not be the right selection.
• Data and Code Portability: When you are deciding what cloud platform to select, you’re not just making that decision for yourself, you’re making it for your customers.  So choosing a platform that doesn’t lock you in to a proprietary codebase or where extracting the data is more of a challenge must be a primary consideration.  Nobody wants to be locked in.  Call it the ‘Hotel California’ effect.  Many companies are wary of the Force.com platform for this very reason, unless they’re building their product in order to take advantage of the Salesforce.com ecosystem.  Also, what kind of protections are you afforded via code escrow?  Think about the challenges that companies who built their businesses — don’t just think about building applications — on Coghead?  For many this was extremely challenging to their business and to some it was fatal.  there’s a financial stability aspect to this as well, so advantage to the mega-vendors like Microsoft, Amazon and Salesforce.
• What Does Existing Code Look Like?: Let’s start out simply: do you have existing code?  If not your choices are much wider.  But if you’re heavily invested in .NET or Java, your choices may be clearer, because the migration path afforded to you will be faster.  And speed does count for a lot.  Here’s one area where Microsoft Azure will have a strong value proposition to existing Microsoft shops.
• Integration: According to Forrester Research, integration is a top concern of clients when selecting SaaS companies.  So does the platform you’re selecting make this challenge any easier on you?  With Force.com, AWS and OpSource Connect you have a lot of existing connectors and modules sitting at the ready that make solving the integration problem easier and significantly reduce the associated coding effort.  Of course there are integration platforms like Boomi and Pervasive Software that you can integrate into your application, but if all else is equal, why not go with the platform that has the integration built in?

What else did I miss?  Please let me know.