Cloud computing: Whose crystal ball is correct?
- 03 August, 2010 03:55
There has been a spate of interesting - and contradictory - stories and research reports about how enterprises are or will embrace cloud computing, topped off by a report by Forrester analyst James Staten that concludes that for all the talk about private clouds, they're a pipe dream for all but a few enterprises.
Kicking off the parade of news was a survey sponsored by Savvis. The survey asked 600 IT and business decision makers about their plans for cloud computing. The results showed that 68 percent of those surveyed thought that cloud computing could help organizations recover from the current economic downturn; respondents expected cloud computing to decrease IT budgets by 15 percent. A full 70 percent of IT decision makers are using or expect to be using an enterprise-class cloud within two years, the Savvis survey said.
Interestingly, 76 percent of survey takers who fell into the category of business decision makers see the "lack of access to IT capacity" as a barrier to business progress. I addressed the way that cloud computing can aid business agility in my blog last week and noted that this type of agility would, ultimately, be the most crucial business-oriented deliverable of the cloud. I also provided a link to our checklist (registration required) to evaluate if cloud computing could aid your business agility.
Given that most IT funding flows from business units, this implies to me that the ultimate beneficiaries of cloud computing (business units and their applications) are eager for quicker access to compute resources as a way of helping them solve their business problems.
Next up was an article in InformationWeek titled "The Why and How of Private Clouds." IW performed its own survey and found that 58 percent of its survey takers are either currently using or plan to use a private cloud. The primary reason IT groups aren't implementing a private cloud today is "no business need," this survey said. Cited as an example is Indiana University, which provides a portal for users to request IT resources, but then provisions them manually, making them available within half a day. Among the survey respondents, the amount of money devoted to implementing a private cloud varies, with over one-third (38 percent) setting aside no specific budget for cloud implementation (15 percent say they devote over 20 percent of their budget to implementing private clouds, with another third saying they spend 10 percent or less).
Then I turned to an interview in Forbes with Mark McDonald of the Gartner Group (Gartner had also done its own survey on the topic of cloud computing as well). This article was really fascinating: Gartner divides IT groups into three buckets: rich, poor, and stuck in the middle.
Rich IT groups (22 percent) have more budget, but crucially, also have more stature within their company. Not only do they get more money to pursue modernization, but also are supported in innovation initiatives. Their interest in cloud computing is more than just cost-cutting; they see it as a way to develop new applications and deliver more strategic value to line business units (I discussed an example of one of these type of applications in my HyperStratus blog, which you can read here).
Poor groups (50 percent), by contrast, don't have much budget and focus on "saving the company's money." In other words, managing efficiently rather than exploring innovation. McDonald paints a bleak picture for these IT organizations, predicting that they'll fade into "administrative irrelevance." He goes on to echo the importance of agility:
"Yes, and the biggest single determinant of effectiveness now is speed. Richer IT organizations can tell you how they raised productivity across the enterprise. Poorer organizations can't tell you that."
McDonald's perspective on cloud computing appears to be directly contradictory to the expectations of respondents to the Savvis survey. In that survey, two-thirds of respondents expected that cloud computing's main benefit would be to save them money; McDonald believes that expectation will lead to being sidelined within the overall company, since it evinces an attitude that IT's major responsibility is to provide commodity services at the lowest possible cost. If nothing else, all these surveys convince one that every survey's results should be taken with a grain of salt, since they all seem to draw different conclusions from a population that should be relatively homogenous.
"You want the truth? You can't handle the truth!"
The famous Jack Nicholson line from A Few Good Men came to mind while reading the capper of the week, Forrester analyst James Staten's report on the prospects for enterprise private clouds. The report's conclusions are, in effect, a bracing bucket of water thrown onto the easy presumption that private clouds are necessary and easy to implement. The report indicates that "only about 5 percent of enterprises have enough experience [to implement private cloud computing]". (italics added).
Forrester identifies four stages of virtualization use:
1. Acclimation: Getting comfortable with the concept and tactical deployments
2. Strategic consolidation: Widespread use and painful transition from virtual server sprawl to virtual server life cycle management
3. Process improvement: Improved utilization rates and further general improvement hampered by functional silos (e.g., separate network, storage groups)
4. Pooling and automation: Use of automated resource management tools, chargeback/utility tracking, and SLA/QoS focus
Only when stage four is achieved is the organization ready to implement a private cloud.
What the report highlights is that private clouds are not solely a technical implementation. They require IT process re-engineering, focusing the organization on end-to-end streamlining, with all manual interactions removed from provisioning and administration. Unless the technical improvements are married to process improvement, you don't have a private cloud. As an aside, the report specifically states that cloud computing requires end user self-service, thereby indicating that Indiana University's arrangement may be an improvement over past practices, but it's not a private cloud, even if that's what IU calls it. The report also implies that those organizations that think they're implementing a private cloud without any investment are living a pipe dream; getting to automation and implementing process re-engineering is many things, but free is not one of them.
Process improvement (or process re-engineering as it's often called) requires more than an updated OmniGraffle flowchart. Process improvement usually implies organization change or restructuring, with some groups being forced to accept a less important role in day-to-day operations or even, in some circumstances, being subsumed into other groups, with its function becoming part of a larger offering by the surviving group.
The term is a bloodless phrase which obscures what it really means: some groups become less important and some groups become more important. Budgets--and careers--are affected as the re-engineering works its way through implementation. Forrester estimates that moving from stage one to stage four of virtualization maturity is usually a five to seven year journey--and recognizing the changes that are required makes it easy to understand why.
Yikes! That timescale doesn't seem like it's really aligned with the expectations or demands that end user business units are pushing in their thirst for greater business agility, does it? A lengthy roadmap with a pot of private cloud gold at the end of it is not going to cut it in today's economy.
Three Practical Steps
So what should you do? Here is what we recommend:
Swallow your pride. It's tempting to assume that you can do it all: "Cloud computing? Of course it's important. That's why we're creating our plan." If you don't have the time to get to stage four, or you fall into Gartner's "poor" IT group, acknowledge it, and start to develop Plan B that doesn't require a big budget boost and a long implementation cycle.
Carve out a smaller cloud within your data center. Maybe you can't re-engineer your entire organization's process and you can't afford to do the entire infrastructure, but perhaps there is a smaller initiative that you can achieve. Focus on a few high-impact apps, or put new apps into the cloud portion of the data center. Just don't, for heaven's sakes, build an internal cloud solely focused on your developers. I addressed this approach last week and called it "type one" cloud agility. Directing your cloud investment toward internal IT optimization is a sure path to career redirection.
Embrace a hybrid cloud approach. Instead of trying to restructure your data center, use a public provider for your cloud efforts. Frankly, I've never understood the mania for implementing an internal cloud. Most of your data center supports existing apps, which don't particularly require a cloud environment, so why implement a cloud when the apps that run on it can't take advantage of it? Instead of sinking a lot of capital into your own data center, why not use someone else's and just pay operating expenses to them?
My own conclusion from the different surveys and articles is that private cloud is an ambiguous term being used indiscriminately during a period of turmoil. The underlying changes in IT--the explosion of applications and data--make the old ways of doing things, with their manual steps and organizational silos, obsolete and wasteful.
Meanwhile, cloud computing offers a vision of far cheaper, far faster, much more agile computing. Unfortunately, the organizations chartered with providing IT services all too often reflexively assume that the new mode of operation must be grafted onto the organizational assumptions and processes that already exist, rather than being viewed as an opportunity to excise waste and inefficiency, thereby marrying compute efficiency with operational effectiveness.
I think we're twelve to twenty-four months away from today's turmoil settling down enough for the real opportunities, requirements, and challenges of private cloud computing to become evident. Until then, we can look forward to a myriad of puzzling and contradictory articles, surveys, and reports!
Bernard Golden is CEO of consulting firm HyperStratus, which specializes in virtualization, cloud computing and related issues. He is also the author of "Virtualization for Dummies," the best-selling book on virtualization to date.