Ignore the jargon. The latest IT buzz really offers the difference between custom pricing and commodity pricing.
By John Sumser
Cloud computing is the IT department jargon bit of the year. The term is widely used to describe computing that happens somewhere else (as in “we do that in the cloud” or “that application lives in the cloud.”) As marketing operations jump on the bandwagon, the distinctions grow vaguer.
It is much less expensive to claim a new feature (or absorb a new buzzword) than it is to implement it. The “me-too” marketing that is predominant in HR software settings almost rewards the distinction blurring of new jargon. While cloud computing might have had a clear meaning at one time, it’s now just another “Heinz 57” blend of non-sequitors.
So, what’s the fuss all about?
A datacenter is a facility full of computing hardware. Lots of enterprise software suppliers sell datacenter services along with their application licenses and implementation projects. If you are running large installations of software, a datacenter resides in the background. It’s where the software and data live.
Imagine, if you will, that there is a scale describing datacenters. At one end is a vendor-specific suite of hardware and software managed by a provider. IBM runs these sorts of datacenters in conjunction with many of their service offerings (and as a standalone offering as well.)
At the midpoint of the scale are datacenters that optimize their offerings based on best-of-breed decisions. In contrast to the first group, these suppliers claim vendor neutrality and a theoretical cost advantage. EDS, among others, offers this type of service.
Both of these datacenter models are sold in contracts that are laboriously negotiated with clients. Details of demand and supply are forecast and argued. Both types of provider work to help their clients find success by building room into the contracts for growth and surges. A successful endeavor is one in which there are no surprises and therefore no surprise outages.
It is an environment in which the operational risk is borne by the customer. If demand exceeds supply, it will result in additional charges and, if the overage is severe, penalties. The supplier designs a solution to a specific customer’s need much like an architect designs a custom building.
Both conventional datacenter types are designed to mitigate risk for specific customers. They are built to be defended. They are smart agreements between vendor and customer designed to meet the unique needs of the customer.
Cloud computing refers (or at least did until the term was hijacked) to a kind of datacenter that is built to be sold. It is the far end of the scale. That they are built to be sold drives the fundamental differences between cloud computing and other datacenters.
This might seem like a subtle distinction, but it’s the difference between custom pricing and commodity pricing. In the first two types of datacenter, the customer bears most risk. In the cloud model, the customer presents demands, and the supplier delivers. It’s more like a utility than an in-house power generation system.
When the provider shoulders the performance risk without a complex specification (i.e. plug and play), all of the over-specifications associated with minimizing the surprises embedded in the contract are replaced by a clear ability to see cost cutting opportunities.
That’s the second difference. In a datacenter, you do cost containment. In a cloud setting, you do cost cutting. In the former, costs are squeezed with reluctance. In the later, cost-effectiveness is central to ongoing profitability. A cloud environment features a relentless focus on profitability.
It’s not too farfetched to say that “cloud computing” is a pricing innovation.
Finding the right solution is always a case of making trade-offs. Datacenters offer a full range of performance and risk control with accompanying costs. Cloud computing is designed to be lower cost, always available, and void of the controls that make datacenters work.