% fortune -ae paul murphy

An interpretation of the cloud

Here's a simple question: from the perspective of a a senior, non IT, executive in a mid range company how does cloud computing differ from the 1980s idea of connecting all users to one big machine (real or virtual)?

Sign up for cloud computing and for each application your people manage one account, interact with one account manager, pay one set of bills, give their users one address to connect to, crash when your vendor or network crashes, and become captive to that vendor's costs.

Sign up for the one big machine model and for each each application your people manage one relationship, interact with one IT manager, pay one set of bills, give their users one address to connect to, crash when your big machine or the local network crashes, and become captive to internal IT costs.

The only real differences such an executive would see, in other words, are in the risks and opportunities that go with each choice - and these divide into two groups:

1. the first group revolves on the idea that it's easier to manage a contract than people - that replacing your time share supplier is easier than replacing your colleague the IT director.

2. the second group consists of differential risk/opportunity tradeoffs. Outside services, for example, limit cost escalation while distributing some failure risks, but take control of corporate data, limit flexibility with respect to change, and impose many more points of failure.

Since, in reality, contracts have less flexibility than people and the trade-offs almost all favor the one big machine model, you have to ask why we abandoned it in the first place - and the answer, I think, is that most of us switched away from the one big machine model because the Wintel revolution against data processing bundled all big systems solutions into the same bag with the 1970s IBM data center, marked that with the skull and crossbones, and then posted "keep away" signs festooned with flashing dollar signs around it.

Today, however, the wintel revolution has failed its users as data centers in larger businesses have come to be every bit as arrogant, unresponsive, controlling, and expensive as their 1970s System 370 predecessors - largely, of course, because the same management ideas that dominated the mainframe era are becoming dominant in the windows one.

So if small machine computing has failed dismally in larger businesses while cloud computing differs from big machine solutions only in that it limits opportunities while incurring more risk, why aren't we all looking at that big machine model again?

I think we are: I think that user interest in cloud computing demonstrates management's unhappiness with the results they're getting from the traditional data center and that the return of the big machine strategy, this time "under new management" is correspondingly inevitable.