Over on the salesforce.com blog, Peter Coffee, Director of Platform Research at salesforce posted in light of the recent problems suffered by Gmail users. Coffee points out correctly that Gmail did not, in fact, go down. Rather users of the web application were unable to access their accounts. As Coffe said;
Gmail did not go down. If you were using an IMAP client such as an iPhone, or if you prefer the simplicity and diversity of using a POP interface, Gmail remained accessible by either of those means. People who assessed the risks of even an occasional interruption of access to cloud-based mail, and found those risks unacceptable, have always had — and will continue to have — many options for providing as much redundancy as they need for as much as they’re willing to spend.
The thrust of what Coffee said then is that the client used to access a cloud capability is only a skin. The capability itself is cloud computing – regardless of how that capability is delivered to, and consumed by, the customer.
While on a literal level what Coffee says is correct, I have to say that it feels like he’s missed the point. Of the millions of Gmail users affected by the recent problems, I’d guess that only a very small proportion actually understand the distinction between Gmail being down and merely the Gmail web app being inaccessible.
Coffee makes the point that it’s over two years since salesforce crossed the point where more than half of their workload is driven by API requests rather than direct access of the website. This may be the case but similarly most users seeing inaccessibility of a salesforce data-driven application would be hard pressed to determine exactly what the issue – salesforce itself, the API or the third party application.
Coffee eventually swings around to what is the bottom line when discussion cloud computing. He contends;
Can you make on-premise solutions arbitrarily reliable? Well, you can always spend more money. Must you give up that freedom of choice in the cloud? Emphatically not — but in the long run, any given level of information security and operational assurance will inevitably wind up costing less in the cloud.
All of this bought to mind something that Vinnie Marchandani bought up recently when he contended that “the best UI is no UI”. Vinnie was riffing on a post by James Governor who was discussing the major portal re-skinning projects taken on by some of the big corporate in an attempt to “make existing enterprise applications and their portal front ends less painful for users” – In his long and compelling post, Governor call for a greater degree of “plasticity”. As he says;
Web 2.0 is more about information than presentation, and the social aspects of information sharing at that – and that’s where plasticity is required.
And of course he’s right – who cares about the presentation… it’s all about the information… kind of…
Which swings us back to the start of this post… yes it is true that users have no real understanding, or indeed care greatly to where the data they are utilizing is stored or manipulated but, notwithstanding this fact, any time that data is unavailable, all vendors in the chain pay the price. And that’s a paradox that looks to continue for a long while yet…