Picking up where we left off in Part 1 of this cloud trends series, let's get out the crystal ball once again for six more predictions on what will happen to clouds in the coming year.
Cloud Trend No. 6: Here Comes One Really Messy Leak.
There was a dramatic rise in the number of publicized security breaches in 2011. Whether this was simply a matter of better detection, increased reporting, or a rise in actual attacks is still a subject for discussion, but one thing that has changed is the attack surface.
Hackers are no longer satisfied to shoulder-surf and socially engineer their way into a single organization. Rather, they're finding their way into public computing systems and using what they discover to launch more targeted, more precise, and harder-to-detect attacks.
Consider reports Bloomberg recently made of a breach at a large hospitality network provider. While the provider has since denied the breach occurred, it underscores a key challenge of the cloud: Sometimes the target is just a means to an end. Far more valuable than the company itself is the list of passwords and clients that can be collected if an intermediate network is compromised. In the cloud, a single hack of an intermediary can yield huge rewards far beyond those of the initial target.
Now consider the launch of an exciting-sounding new Web application, run in the cloud, paid for with Vanilla Visa and a throwaway address. Thousands of people sign up for the beta with their email and a password. We know that the vast majority of people reuse their passwords across sites, so it's easy to harvest likely passwords with minimal effort.
Something's got to give. And in 2012, it will. We'll see at least one spectacularly messy leak and, of course, we'll blame the cloud. After all, that's what made the attack possible. And that's where the breach happened. Rather than focus on the problems behind the attack--zero-day exploits, poor password behavior, naive mobile users--we'll once again debate whether clouds are secure.
This time, however, the breadth and impact of the leak will be enough to spur regulators into action, enforcing two-factor authentication (using mobile phones' location awareness alongside passwords) and password generation and refreshing policies.
Cloud Trend No. 5: Consumer Expectations Come Back To Roost.
Today's workers come to the office knowing more about what their friends had for dinner than about whether the company will make its sales targets or where the truck fleet is located. Raised on a diet of feeds, friends, likes, and retweets, they have a set of expectations about transparency and real-time information that enterprise IT simply can't satisfy.
We've talked about enterprise collaboration for a long time, but those expectations are finally starting to provoke change. One of three things can happen.
-- IT can set up its own enterprise collaboration systems to retain control. Given cost and timing constraints, these will likely run in the public cloud, often as software-as-a-service tools or integrated into existing, approved products that have already cleared the company's "IT immune system."
-- IT can implement tools to regulate the use of public systems. Bolting compliance, auditing, single-sign-on, and account management onto consumer platforms like Twitter and WordPress isn't easy, however, so some companies will simply turn a blind eye.
-- It can do nothing. Then it will take a visible breach or outage of a system it's relying on to force the company to take action. Given the increasing dependency of organizations on "back channel" communications and a rise in commercial hacking, this is likely to happen.
We gave IT a monopoly on technology in order to complete great projects, much as we created monopolies to build transcontinental railroads or put phone lines in the ground. But just as trains and telecom giants outlived their usefulness, turning innovation into complacency, so enterprise IT no longer has a guaranteed monopoly.
End users are consumers--and so are executives and shareholders. The expectations they bring to the office and boardroom are set by public clouds and technology upstarts. IT will have to satisfy a significant number of these expectations in 2012, or face the wrath of its stakeholders.
Cloud Trend No. 4: Spikiness.
When the first multi-core processors came out, they weren't well used. Code could use only one processor, and the other sat idle. It took time for software to catch up with hardware, but eventually compilers took advantage of multi-processor computers, finding ways to execute tasks in parallel so they'd run faster.
This pattern is happening again, only this time, instead of multi-core chips, it's multi-machine clouds. Today's code has to be specially engineered to take advantage of a cloud--that's what frameworks like Hadoop do. But it's only a matter of time before compilers get an "optimize for cloud" checkbox, and before the code they generate is smart enough to grab all available computing.
If you're a capacity planner, that scenario should terrify you. One of the oldest tenets in cloud computing is that one should "own the base and rent the spike." What happens when the majority of applications can burst to consume all machines in the data center? For one thing, utilization varies far more; for another, these spikes happen more frequently.
That means we need new planning tools to measure and constrain these newly elastic applications. It also requires management technology that can optimize the tradeoff between faster results (at the expense of the entire data center or a big bill from the public cloud) and frugal execution (in return for delayed answers). In 2012, we'll see the emergence of management tools that consider user experience and provider cost, and compilers that can sprawl their workloads opportunistically.
Cloud Trend No. 3: The Advent Of Big Data.
Big Data is perhaps the only term in IT less well defined than cloud computing. The rise of large, unstructured datasets, along with the tools to crunch and interact with them, is poised to change every facet of how organizations make decisions. Big Data requires big systems to work, and those are nearly always cloud-based unless you're Google, Facebook, or Twitter.