“Our No. 1 ambition for cloud computing is economic,” says Johan Goossens, head of NATO Allied Command Transformation’s Technology & Human Factors Branch in Norfolk, Va.
“With reduced budgets everywhere in the world and certainly in defense spending, we need to consolidate IT operations. And we think that the cloud offers us a chance to achieve some of that consolidation while improving information flows [among the 28 NATO nations] and allowing us to save some money,” he says.
Toward that end, ACT is evaluating a number of cloud projects, and expects to find value in the private cloud as well as in public cloud infrastructure, platform and software services. “I’m convinced there will be savings, but I haven’t quantified them yet,” Goossens says.
Therein lies the rub.
Depending on your situation, cloud computing might not actually save you money – and even if you’re convinced that cloud is saving you money, proving it can be difficult, says Geva Perry, an independent cloud strategist and author of the “Thinking Out Cloud” blog.
“When you see statistics like moving to the cloud can save you roughly 10%, you know those calculations are squishy,” says Dave Hart, CTO of Presidio Networked Solutions, an IT services company. “All this economic stuff is in the eyes of the beholder.”
For one, comparing the cost of using a public cloud service to that of buying and running hardware and software is an apples-to-oranges exercise.
Trying to factor in every little element is tough, Perry says.
“You’ve got to do things like figure out how much you’ll pay for per-use bandwidth as opposed to what you’ll pay when you strike a bandwidth deal. And, you’ve got to take into account the cost of maintenance and management of the IT infrastructure, including what’s outsourced and what’s not and what percentage of IT headcount goes away with a move to the cloud, for example,” Perry says.
One of the only ways an enterprise can get a truer understanding of the costs is to start running applications in the cloud and then calculate the actual spend, Perry advises.
But that raises a slew of additional issues, such as how to account for fluctuating prices characteristic of an early market.
“If you look at Amazon, you’ll see that it’s had something like two dozen pricing announcements in the last two years – so the calculations you’ve made two weeks ago may no longer be relevant because your cloud provider has just slashed prices on one of its services you use,” Perry says.
Also be aware that the pricing you pay for that experimentation won’t necessarily mirror the cost of a production deployment.
If you’re experimenting with Amazon’s Elastic Compute Cloud, for example, you’d most likely pick and choose the resources you need from among the provider’s On-Demand Instances, Perry says. These let you pay for compute capacity by the hour with no long-term commitments.
But when running production applications in the cloud, you’re more likely to use Reserved Instances, which give you the option to make a one-time payment for each reserved instance and in turn receive a discount on the hourly usage charge for that instance, according to Amazon pricing charts.
Will Gregerson, CFO of Schaeffer Manufacturing, a specialized lubricant manufacturer in St. Louis, says he faced similar frustration in running price comparisons between traditional ERP software and a cloud ERP offering.
One ERP vendor offered a flat site license, another added in costs for extra modules required. The ultimate decision came down to SAP Business One traditional ERP software and NetSuite’s cloud ERP service, he says. SAP charged a per-user license fee – which would be expensive to start up, he says – while NetSuite also charged a per-user fee, but in a pay-as-you go model.
Ultimately, money was not the deciding factor; rather it had to do with ease-of-use and functionality. “The pay-as-you-go model provides a nice benefit, but we would have written the check for the SAP Business One if we had decided that’s what we wanted.”
Schaeffer opted for the NetSuite cloud ERP because, “We let somebody else do all the programming and we get a much better system,” he says.
Furthermore, Gregerson’s analysis was that the crossover point at which the cost for the pay-as-you-go cloud ERP would finally exceed the upfront cost for the SAP software would be nine or 10 years away – too far out to be a serious factor in the cost consideration, Gregerson says.
It’s not all dollars and cents
For many enterprises, factoring in soft savings is yet another challenge.
“How do you measure the cost of agility?” Perry says. “I’ve just allowed my developers to get this new product or service up and running two weeks faster than they could before and they can make changes very easily, so how much money is that worth to me? Most good executives have an intuitive feel about how the cloud is what they should be doing, but it’s hard to quantify.”
Often when a cost analysis shows increased spending on cloud, an enterprise IT executive will proceed with the project regardless, Perry and other cloud watchers say. “We’re seeing more and more success stories where the soft benefits like business agility are proving more important than raw cost savings of the hardware infrastructure,” Perry says.
Proving cloud economics can be rather nebulous even when comparing an internal private cloud to traditional IT infrastructure, agrees James Staten, vice president and principal analyst with Forrester Research.
“When we examine cloud customers, what we find is that they’re more efficient, they’re faster and their overall IT operations cost for the cloud is way lower than it is for their traditional environment – so the benefits are clearly there. But it comes down to this ability to calculate the intangibles,” he says.
As an example, he cites the case of one company that built an internal cloud but didn’t reduce IT headcount at all. “They have exactly the same number of people, but those people are handling three times the number of virtual machines in production. So that’s a definite operational advantage for the company, but one that can’t be turned into a dollar value,” Staten says.
On the other hand, the company might calculate a dollar cost-avoidance, he adds. “We had this ratio of administrators to virtual machines before and this many admins to virtual machines now. Before, we would have needed to hire five more administrators to handle that number of virtual machines,” he says.
And while Forrester has seen evidence that IT operational costs run lower for internal clouds vs. traditional infrastructure, it’s not ready to pinpoint a specific percentage savings. “There just aren’t enough examples yet,” Staten says.
Instead, Forrester cites its virtualization maturity model. “We talk about four stages of virtualization, with the last stage pretty much being like a cloud operation, and we see roughly a 10% to 20% operational efficiency improvement per stage.”
Certainly in the pilot and “toe in the water” stages, Deloitte Consulting finds that most clients’ cloud business cases have been positive, says Mark White, CTO of the firm’s technology practice.
But the firm has seen clients make the decision to forego the cloud, at least for now. “There have been cases where a CIO has said, ‘This just doesn’t make sense. I have critical mass of skills and scale, the current solution is working well, and it makes the most sense for me to continue on that path with an upgrade, enhancement or refresh,” he says.
The important thing is that they’ve made this decision after analyzing the business case, White adds.
“Cloud is not something around which you should roll the dice. Do the business case. Do the analysis. You’re not going to be spending this money for the sake of technology,” he says. “It’s cool, don’t get me wrong. As a CTO I think it’s really cool, but that’s not a reason to spend money on the business.”