A year ago I laid out my predictions for cloud computing in 2011. In the spirit of honesty and willingness to display my errors in public, I thought it would be useful to recap those predictions, grade them, comment on them and review an exciting, tumultuous year.
Below are my five predictions for cloud service providers and five for end users, followed by comments and a grade for each one.
Cloud Service Providers Predictions
Prediction 1: The CSP business explodes ... and then implodes CSPs will continue to pour money into building cloud computing offerings. Large companies will invest billions of dollars constructing data centers, buying machines and infrastructure, implementing software platforms, and marketing and selling cloud services. Regional and local players will likewise do the same, albeit on a smaller scale.
Comment: Essentially, I noted that being a CSP is a capital-intensive, complex business, and after sinking money into building a cloud offering, many players would recognize that their me-too, undifferentiated offerings were never going to be able to compete with the largest, best-funded and most technically sophisticated CSPs. Once the smaller players realized this, they would throw in the towel and look to be acquired.
The counter-argument is that many providers, despite their lack of size, will offer customized service, industry knowledge, and, possibly, a willingness to live on thinner margins. After all, there are numerous small hosting companies -- why would cloud computing service providers be any different?
Clearly, I was too early on this one. Far from a consolidating market, as we reached the end of 2011, more companies were announcing that they, too, were getting into the CSP business. I still stand by this prediction, long-term, and perhaps last week's announcement is a harbinger of the future. Maybe small players can thrive on service, but I think cloud computing represents a step change in infrastructure, akin to mass production, and inefficient providers will be squeezed out -- eventually.
Prediction 2: Market Segmentation via Customer Self-Selection Many vendors and commentators feel that the SMB market is a natural for infrastructure-as-a-service (IaaS) computing because of their lack of large, highly skilled IT staffs. Sometime next year everyone will realize that removing tin still leaves plenty of challenging IT problems, and cloud computing delivers a few new problems besides. Once that realization sinks in, everyone will agree that SMB are a natural fit for SaaS and that only larger companies should imagine themselves as IaaS users. Consequently, SaaS providers will gain an even higher profile as adoption rates increase. However, SaaS will by no means be only an SMB phenomenon -- far from it. SaaS will become the default choice for organizations of all sizes that wish to squeeze costs on non-core applications.
Comment: Got this one right, I think. I hear much less of the "IaaS is great for small business" discussion today and SaaS is on fire. The Q4 acquisitions of SuccessFactors and RightNow by SAP and Oracle, respectively, illustrate that even the largest packaged software vendors can read the handwriting on the wall.
Prediction 3: OpenStack will come into its own The attractiveness of a complete open source cloud computing software stack will become clear and interest and adoption worldwide of OpenStack will grow during the next year.
Comment: Chalk another one up as right on the money. Many leading tech vendors such as Cisco, Dell, HP and Citrix have adopted OpenStack. Many end users are evaluating it as a potential private cloud infrastructure foundation. And it has a lot of experimentation going on in the academic and individual user community. In addition, at least two companies, Nebula and Piston, have received venture funding to create spin off distros of the OpenStack open source code base.
OpenStack now confronts the challenge of living up to its buzz. Linux, to which OpenStack is often compared, was around for five or six years before it got big in the industry. That gave it time to stabilize and learn from real-world feedback. OpenStack doesn't have that luxury. This year is critical to see the buzz turn into results.
Prediction 4: Cloud computing takes off in emerging economies Much of the angst about what form of cloud computing end-user organizations should use (see End User Predictions below) doesn't exist in emerging economies. Most companies have no significant installed base of infrastructure, so the urge to repurpose existing hardware (or, more realistically, avoid prematurely writing off underappreciated assets) is not relevant.
A good analogy for what is going to occur with cloud computing in emerging economies is what happened with telephony. Most of these countries, as their economies developed, hopscotched right over fixed-line telephony, and moved directly to mobile as the primary form of telephony, based on its convenience, flexibility and lower cost.
Likewise, we'll see a rush to cloud computing since it does not require significant end-user investment in wasting assets. Don't be surprised if the growth rates of cloud computing in emerging economies far outstrips that in more developed nations.
Comment: I would give this a mixed grade. Clearly, there is tremendous interest in cloud computing in emerging economies. This year, we worked with a small South American country that is seeking to spread technology use by reducing the barriers to adoption, with enormous potential for their economy and society. Amazon established its first non-developed economy location in Brazil, indicating that the largest provider sees opportunity in growing economies.
On the other hand, you have to evaluate announced plans against actual adoption. I can't tell you how many cloud infrastructure software providers have shared with me that they "dominate the mainland Chinese market." They can't all be right, and when pressed for numbers of launches and adoption statistics, they quickly turn to querying whether one wants another sweet roll.
As an overarching comment, the issue of customer adoption will become much more visible this year. Amazon is clearly experiencing huge growth, but when you move past the biggest couple of players, the actual growth is less verifiable. Certainly, we've seen many end users struggling with how to make sense of what they should do now that they're convinced they should do something about cloud computing. Lydia Leong's (Gartner cloud analyst) comment that she works with many companies that are desperately trying to identify the smallest possible project they can execute to respond to upper management's "get on the cloud" directive indicates that wholesale adoption is not underway.
This year we will hear many stories of users grappling with how to architect, implement, integrate and manage cloud applications. This will illustrate the cloud adoption gap vividly.
Grade B, with some makeup work to come
Prediction 5: Continued rapid innovation by CSPs and SaaS companies Many people point to the astonishing rapidity with which AWS continues to roll out new features and service offerings. Its launch this week of Route53, a robust and inexpensive DNS service, is just one example of the company's continued innovation. However, AWS is by no means alone regarding innovation and creativity. Next year we will continue to see amazing new offerings from companies that create services based upon cheap and scalable infrastructure.
Comment. Think I nailed this one. The PaaS offerings from both startups and established vendors like Red Hat and VMware reflect tremendous innovation. Writing elastic IaaS apps is not trivial, and making it easier will drive adoption (see last prediction). This trend is still nascent, but it holds tremendous promise and one can easily foresee a rosy future for PaaS.
Moreover, AWS continues to innovate and improve its offerings in both small and large ways, and their rapid innovation threatens to leave other providers in the dust. We've been struck by how many providers operate more like MSPs than automated cloud providers. Merely parroting claims of "enterprise class" isn't enough to achieve success, particularly when anything beyond the basics requires expensive professional services. The bedrock of "enterprise" is "I want it my way" and making it hard to do anything but the vanilla offering seems self-defeating.
Turning to SaaS, it's obvious that this mode of standardized offering is the clear future for packaged software and the rapid growth (and earlier-mentioned acquisition) of SaaS illustrates this dramatically. Why should every user have to undergo the heartburn of implementation when 99 percent of them do exactly the same thing?
We've been struck by how many clients are willing to take on the work of migrating to an online alternative to existing on-premise product. They calculate that it's better to pay the price once than undergo the repeated tortures of upgrades at regular intervals. Hats off to Microsoft (which has more to gain -- or lose -- from this transition than anyone) for embracing it and aggressively rolling out a suite of office automation products.
End Users Predictions
Prediction 1: Focus on cost and transparency I admit it, I carry an economics bias. While many people point to the undoubted advantages of cloud computing -- agility, elasticity, self-service -- my perspective is that the cloud computing revolution is that these characteristics are not new or even specific to cloud computing.
The revolution of cloud computing is that, due to scale and automation, those characteristics, for the first time, can be achieved at an attractive price point -- which makes all the difference. Moreover, they are delivered by CSPs in a completely transparent manner -- listed in black and white on the provider website.
What this economic revolution and cost transparency will translate to in 2011 is a demand that internal IT groups provide the same level of transparency, and woe betide the CIO who proffers feeble excuses like "we're not really set up to identify specific costs"
Comment: I think I was wrong, or at least way premature. In our work, we've seen some companies planning internal clouds who assert they can beat public provider costs, but when we drill down into their assumptions there's a lot of hand-waving that "proves" the internal option is less expensive.
We've seen other IT organizations that assert they'll offer "showback," illustrating what resources are used by application. Frankly, this seems like a losing game. Application groups don't care how much resource they use; they care whether their application is working properly, and if the IT organization has to spend a ton of money to achieve that, so what? Only when resource consumption is tied to cost will we see real application attention to consumption efficiency.
And we've definitely seen application groups more demanding of price transparency and a desire for consumption-based costs. There's even a bit of exasperation along the lines of "XYZ cloud provider can tell me what everything costs, why can't you?"
Longer term, I believe this trend is unstoppable. Opaque costs are the realm of bureaucracies and monopoly providers (i.e., your phone bill). When convenient, transparent, cheaper alternatives exist, demand for better pricing models will be irresistible.
Prediction 2: More public/private cloud confusion The current debate raging about the right mode of cloud use will continue, unabated and may even get worse. There are valid arguments for both options and I won't go into them here, as I've addressed the topic several times, most recently here. However, it can be said with confidence that the pressure to provide some cloud option is only going to increase.
The desire by application groups and business units to grasp the benefits of cloud computing is palpable; one example surfaced when the U.S. Federal government announced that, starting in 2012, federal agencies are being told to default to cloud-based solutions "whenever a secure, reliable, cost-effective cloud option exists." What this means is that, if you're a CIO, whatever form you want to deliver cloud computing in, it had better be ready in 2011. Extended rollout plans based on lengthy private cloud initiatives won't cut it in the fervid rush to deploy cloud computing.
Comment: Despite everyone proclaiming that they're tired of this discussion, it continues unabated. Every vendor we've spoken to has said the greatest source of growth in 2011 was companies planning to build a private cloud, so the demand is clearly there for the private variant. The public side is mixed. As I noted earlier, AWS is growing like wildfire. As a side comment, I believe that last week's AWS spot pricing jump indicates that demand is such that Amazon has little spare capacity, indicating that its growth is likely to continue, and probably continue to accelerate. Many of the other public providers, however, have much more mixed growth records.
The confusion and muddle in this area will, if anything, grow during 2012. Many IT groups view building a private cloud as a core requirement and will focus on it throughout the year. Application groups will make their decisions based on what they think is most productive, buffeted by lobbying and sales efforts by all providers, both internal and external.
If you're tired of this discussion, be prepared to be even more fatigued in 2012.
Prediction 3: More hybrid cloud confusion I see more vendor hype and end-user wishful thinking on this topic than any other in cloud computing. Vendors breezily assert and end users blithely repeat that the future will be applications effortlessly, transparently and automatically migrating between internal IT infrastructures and external cloud providers.
Comment: Totally right on this. The vendor display area at CloudExpo was filled to the brim with vendors proclaiming how their solution provides easy and transparent "hybrid cloud functionality." All of them stated or implied that workloads would "automagically" transition back and forth between an on-premise and external cloud.
More astonishingly, when queried about how this magic functionality works, booth personnel either couldn't describe how it operates, or described use cases so circumscribed that they wouldn't address anything one would recognize as real-world. I feel sorry for IT organizations that are bedazzled by the smoke and mirrors and only later realize that they've wasted time (and, perhaps, money) on technology that can't deliver what it promises. Given the rich history of enterprise software vendors over-promising and under-delivering, maybe this is what it meant by enterprise cloud.
Grade: What's higher than A+?
Prediction 4: Application architecture challenges As IT organizations deploy their first cloud computing applications, they'll find achieving agility and elasticity is hard--and requires new application architectures.
Implementing robust applications to run on less-than-robust infrastructure imposes design requirements for redundancy, failover and session isolation. Designing elastic applications that can automagically grow and shrink in response to application load necessitates functionality that allows graceful on-the-fly configuration without human intervention.
Comment: Right, and more to come. Most IT organizations are absorbed in deciding what form (and provider) of cloud computing to adopt and have not yet confronted the challenges of designing applications to run in whatever environment is ultimately selected. However, it's nearly universal that once an IT organization gets around to trying to build those magic elastic apps, they learn the difference between virtualization and cloud computing, and recognize how much additional effort will be required to gain the benefits of cloud computing.
We've worked with companies this year who learn how much is changed by implementing "true" cloud applications. Achieving robustness, disaster recovery, application lifecycle--and more--are affected by running elastic applications in cloud environments. I believe we'll hear much more about this during 2012.
Prediction 5: IT operations challenges Operations will be challenged in three ways during 2011. The first challenge is associated with process re-engineering. The manual operations practices in place at most organizations aren't sufficient for the self-service vision of cloud computing. Application groups will clamor for the immediate resource availability associated with the public cloud providers, and will expect internal IT operations to respond as quickly. That's challenge number one.
Comment: Right, but it's still early days. What many people fail to understand is how quickly applications are growing in complexity and scale. Applications are way beyond the venerable three-tier and often contain multiple levels of caching, calls to external services, different types of databases, and use of one or more content distribution networks. Yet IT organizations, for the most part, struggle to manage this complexity with the same operations processes used when a single web server talking to an Oracle database through a WebLogic application server was the acme of complexity.
It's unfortunate that more advanced practices have picked up the label "DevOps." Not because it's wrong or inappropriate, but because the label attracts skepticism and derision. The issue is how to run applications in production and the fact that the old methods can't keep up. Whether you call it "DevOps" or choose another label, you will confront this in the future.
Cloud Computing Conclusion
On balance, the grades show an inclination toward premature prediction. The ones that graded out poorly reflect a slower pace of adoption than expected or, put less charitably, a misguided belief in how quickly organizational change occurs. While of none of the predictions are flat-out wrong; neither are they completely perspicacious.
What you can say is that, despite many calling cloud computing over-hyped, 2011 saw end-user interest in cloud computing continue to grow dramatically. The predictions' pace may have been wrong, but the direction was, I think, right.
In my next post, I'll take a crack at what 2012 will bring for cloud computing.
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.
Follow Bernard Golden on Twitter @bernardgolden. Follow everything from CIO.com on Twitter @CIOonline
Read more about cloud computing in CIO's Cloud Computing Drilldown.