If crisis equals opportunity, today's economy offers the best prospects for 70 years.
That's good news for disruptive technologies, yet for cloud computing the recession has come just a little too early.
Vendors are telling customers the cloud's not ready for enterprise, and Gartner agrees, saying the technologies will take seven years to mature.
Still, now is the time for IT organizations to be examining the cloud business case. They could do worse than work their way through a paper by boffins at the UC Berkeley Reliable Adaptive Distributed Systems Laboratory (aka the RAD Lab).
The authors, who have even come up with a formula for making the switch, say the cloud is an old idea whose time has come thanks to the massive commodity computing grunt generated by data centers. The cost of "electricity, network bandwidth, operations, software, and hardware" have come down by a "factor of five to seven."
They begin with a helpful definition: cloud computing covers both the software and services delivered remotely and the hardware and systems on which they are hosted.
Using 2006 figures, they calculate that the network costs in a medium-sized data center were $95 per Mbps per month, compared with $13 in a large center. Storage costs were $2.20 per GB per month versus 40 cents.
According to RAD Lab, the cloud is offers the "illusion of infinite computing resources" on-demand. Users don't have to plan far ahead for provisioning, or make up-front hardware commitments. Companies can start small and increase resources as needed, and can pay as they go for computing resources.
Like other IT outsourcing arrangements, enterprises also get access to new technology as it becomes available.
The paper canvasses the potential profitability of the cloud business for providers: at one end is Amazon's EC2, delivering pretty much raw hardware, at the other is Google's AppEngine, which exclusively delivers web apps, with Microsoft's Azure in the middle.
The most interesting aspect for users is the discussion on adoption. The great advantage of the cloud is its elasticity - the "ability to add or remove resources at a fine grain" - which aligns the investment in computing with an organization's needs.
That benefit can only be realized at the right cost points, and the paper steps through a number of buy-or-rent scenarios. But while the hardware costs are generally quantifiable, the software management costs depend on the level of service offered by the provider as well as the complexity of managing on- and off-cloud software systems.
Also useful is a table of obstacles to the development of cloud computing, such as performance predictability, data security, and the proprietary nature of cloud APIs.
There's no rush to implement cloud right now, despite the hype. For the future, however, the paper draws an analogy with the chip sector. "[I]t would be as startling for a new software startup to build its own data center as it would for a hardware startup to build its own fabrication line."