My last couple of columns have addressed cloud adoption patterns by IT organizations. Has Cloud Computing Been A Failed Revolution discussed the seeming ennui regarding cloud computing on the part of IT groups – that they seem less interested in the field, despite the belief on the part of vendors that the cloud represents tomorrow’s technology infrastructure. Most recently, The Real Cloud Computing Revolution described three real-world examples of companies using cloud computing to solve problems they couldn't have addressed in the infrastructure models of traditional IT.
This week, two sets of information brought to mind William Gibson's well-known aphorism "The future is already here – it's just not very evenly distributed." Despite evidence such as that outlined in my most recent blog post, each describes a future of IT as little changed from a decade ago – mostly on-premises, mostly static with little possible resource elasticity and mostly unable to support the kind of innovation described in that post.
In other words, the future of IT is already here, but it's being carried on in areas far removed from the traditional IT settings. The key question: Will those traditional settings ever catch up, or will they be abandoned to become a kind of technology rustbelt reminiscent of the post-industrial outcomes seen in today's Detroit?
Forrester: A Tale of Two Infrastructures
First, a Forrester report. Despite its title, The Public Cloud Market Is Now in Hypergrowth, which implies that public cloud computing is becoming a dominant player, it actually presents an infrastructure message that paraphrases of the opening lines of A Tale of Two Cities: It was the cloudiest of times, it was the most traditional IT of times.
[ Analysis: The Truth About Enterprises and the Public Cloud ]
Forrester surveyed a large number of IT personnel and outlined findings that show that cloud computing is growing rapidly. Over the next five years, SaaS will experience compound annual growth rate of 14 percent, reflecting the fact that it grew so rapidly over the past few years that future growth will be somewhat muted, while IaaS will experience a much stronger 38 percent CAGR.
On the face of it, the theme of the report is that cloud computing is picking up steam. However, delve deeper into the survey numbers and a different message emerges.
As one datapoint, the report notes that SaaS has grown quite rapidly in the recent past, in part by displacing on-premises applications. For the most recent three years, the percentage of companies reporting, for application types such as ERP and CRM, that they have replaced an on-premises application package with a SaaS offering has averaged something like this:
- 2011: 3 to 5 percent
- 2012: 5 to 7 percent
- 2013: 12 to 14 percent
As another datapoint, the report states that public IaaS/PaaS is being used to replace on-premises infrastructure as well as create what Forrester refers to as "complementary applications." These extend on-premises applications to provide more functionality, as when new cloud-based applications extend existing systems of record to enable richer engagement with customers or partners.
[ Related: The New Cloud Application Design Paradigm ]
However, as to the overall future of IT infrastructure, the Forrester report presents a very different picture. Looking out five years, one sees that, according to Forrester's estimates, the vast majority of total server and storage spend won't be put into service in cloud environments. Of 2020's $230 billion total investment, Forrester sees only 19 percent serving as cloud infrastructure. I interpret this to mean that 81 percent will be spent on supporting traditional IT infrastructure – on-premises data centers with manually managed environments running static application topologies.
Overall, the message of Forrester's report is mixed. Public cloud computing is experiencing rapid growth (hypergrowth, even), but that growth represents but a small percentage of overall computing, even five years into the future. Bottom line: The future of IT infrastructure looks a lot like the past, with a smidgen of public cloud computing overlaid on it.
Why the Cloud You Want Isn't the Cloud You Deserve
According to Asay, Perilli notes that many people expect an on-premises cloud to provide the same capabilities as a public cloud: Simple access, consumption-based cost, easy elasticity and low-overhead governance. Instead, they find that most on-premises cloud environments are complex, expensive, costly, heavily governed (presumably, this means highly manual, trouble ticket-based management) and capacity constrained.
But according to Asay, Perilli says this shouldn't be a surprise. The vision that on-premises clouds can easily be implemented is one foisted by vendors onto IT organizations. Instead of being surprised or frustrated by the limited capability of an on-premises cloud, recognize the following:
- It's early in most IT organization's cloud implementation efforts. These efforts should be measured against on-premises existing infrastructure capabilities, not best-of-breed public providers.
- Providing "real" cloud capabilities (simple, cheap, low-overhead governance) requires significant restructuring and process re-engineering. That's hard. Expecting too much too soon is unrealistic, given the constraints under which most IT organizations operate.
- Instead of expecting "real" cloud capabilities today, IT should focus on a long-term improvement process, starting with something achievable – say, a test/development cloud that addresses software engineer desires for agile development – and methodically proceed over the next few years to layer on additional "real" functionality.
Warped Thoughts on Future of IT, Cloud Shouldn't Be Surprising
Looking at the Forrester report and Asay's blog, I'm struck by the conservatism of these two predictions. Essentially, the messages can be boiled down to two realities. First, even five years from now, the vast majority of IT infrastructure will operate much as it is today and as it has been for the past 15 years. Second, IT organizations should accept that as a given and respond with rather ponderous 10-year plans to eventually deliver similar functionality as public providers have for the previous decade.
Set against the kind of examples presented in my last column (cloud adoption by companies such as Lotus Racing, Lonely Planet and Marks & Spencer), and against the heightened expectations of a world rapidly restructuring to become more digital, it seems that these predictions fall far short of what's going to be required of future IT. These company's pressing business challenges require immediate response, not a tepid, half-hearted, leisurely journey toward a future, somewhat more agile infrastructure.
A strategy based on a future that's more or less like today, with plans predicated on a 10-year journey to offering what Amazon and its fellow CSPs provide today, is fraught with danger for IT organizations. It presents the very real potential that IT will be seen as an irrelevant backwater to be bypassed with technology initiatives better suited to a more rapid pace of change.
How can the conventional wisdom of how IT should prepare for the future be so out of step with what's going to be needed? In considering disconnect between today's conventional wisdom and tomorrow's real requirements, there are three primary reasons today's vision is so inadequate.
It's too early. The ultimate outcome isn't obvious. We're around the one-decade mark of cloud computing, and perhaps it's not clear how this will all turn out. For a previous platform revolution, the PC, the equivalent 10-year mark is around 1988. At that time, many in the industry treated PCs as limited "toys" unsuitable for the kind of real computing done by minicomputers. In 1988, Digital Equipment was the No. 2 provider in the industry, with $14 billion in sales; by contrast, Dell did $159 million.
Ten years later, Dell achieved $10 billion in revenue; DEC was a badly tarnished has-been put out of its misery by Compaq's acquisition. Similarly, it's too early to predict how things will really turn out in cloud computing. Perhaps in five or six more years actual results will prove these predictions not aggressive enough.
Our vision is too limited. Asking established practitioners about the future naturally results in a limited vision, since they're bound to a mental framework formed by traditional practices. As an instructive example, the fascinating book The Second Machine Age discusses how, when electric motors began to replace steam power in factories, factory designers continued the practice of a large central motor running manufacturing machines via complex systems of belts and pulleys. They couldn't envision new designs with smaller motors attached to the machines themselves. It took a new generation of factory designers to internalize the potential of smaller electric motors and create designs more appropriate for their capabilities.
[ Analysis: Business Agility Drives Cloud Adoption ]
Likewise, IT personnel raised on expensive, largely hand-managed environments are bound to envision a future that continues established practices, even if the new capabilities associated with virtualization obviate their need. It will require a new generation of IT practitioners to comprehend cloud computing's potential; when they come into power, established systems will be discarded in favor of the new.
We're asking the wrong people. In a fascinating discussion about newspapers, called Newspapers and Thinking the Unthinkable, professor and media analyst Clay Shirkey notes that, in all publishers' discussions about how the Internet would affect the news business, one assumption was unshakeable and unquestioned: Whatever news solution eventually emerged would require large, print-focused new organizations as the primary distribution mechanism. There was no ability to imagine a journalism world that didn't have major media companies such as The New York Times and The Los Angeles Times at its heart.
From the established, dominant provider's perspective, every solution had to continue and reinforce established arrangements. Asking IT about the future of IT infrastructure inevitably results in prescriptions that maintain IT's current position, even if the outcome is unsatisfactory to users. Stated more cynically, asking IT about public cloud computing is like asking the turkey about Thanksgiving – you can be sure of a negative response.
Notwithstanding the reasons why today's cloud predictions are too conservative and likely to fall short of what the IT world of 2020 will actually look like, one fact remains: These prescriptions are completely inadequate to address the requirements of our economy and society circa 2020.
As William Gibson's quote indicates, the future of IT is already here. It can be seen in the examples described in my last column. It's a mistake to see the examples as outliers unrepresentative of real applications. In fact, they represent the real future of IT. The challenge for IT is to prepare for that future rather than a future that reflects nothing more than the linear continuation of the past.