By Leslie Gospill
If you believe the industry hype, companies worldwide flock to the cloud for their IT infrastructure at such a rapid pace that the question isn’t about “if” anymore, but rather, “how fast” migration can take place—and “what” should go first. The barriers to adoption that existed just two years ago hardly get a second thought anymore, and cloud service providers of all sizes and capabilities are maneuvering in the market space to stay relevant.
Current research supports this sense of urgency, as cloud-based infrastructure as a service (IaaS) becomes the “new normal”—and migration seems to become more turnkey. RightScale’s 2016 State of the Cloud Report shows a 40 percent increase in adoption among the 1,060 organizations surveyed worldwide in just the past two years. And Goldman Sachs Research anticipates a 30 percent compound annual growth rate in IaaS spend, noting that we’re really only in about “year six of a 20-plus-year cycle.”
That last statement is the “tell.” Even though it was launched by Amazon in 2007, IaaS is actually still in its infancy. In last spring’s Magic Quadrant for Cloud Infrastructure as a Service, Worldwide, Gartner noted that, “the market for cloud IaaS is in a state of upheaval.”
Richard Seroter, VP of product management for cloud, CenturyLink, agrees. “Smart observers are realizing that this is still a very fluid space,” he says. “And that’s a good thing. But customers are going to have to pay close attention to understand the providers, roadmaps, strategies, and even the fabrication patterns that are evolving on an almost monthly basis.”
The number of cloud providers serving the IaaS market has shrunk a bit over time—as is common in IT, due to acquisitions or changes in business strategies. Currently, the companies to watch include newcomers DigitalOcean and Oracle; niche players Rackspace, CSC, and Joyent; “up-and-comers” CenturyLink, VMware, IBM Softlayer, and Google Compute Engine; and the market-beating giants Microsoft Azure and Amazon Web Services (AWS). Amazon, in fact, boasts ten times more cloud IaaS compute capacity in use than the aggregate total of the rest of the market, according to Gartner.
But as ubiquitous as cloud-based infrastructure may seem, it still represents only approximately three percent of the annual global IT spend. “We’re really only in the first or second inning of a nine inning game,” says Seroter.
“There are some huge, cloud-native market leaders, like Amazon and Google, that have an awesome head start,” he adds, “but I think it’s a very lucrative place for many of us to continue to add value.”
Why the Hype? Drivers to Adoption
IT infrastructure services like compute, storage, and networking defined the first generation of cloud computing adoption. Startups with very little capital were only too happy to take advantage of this metered, on demand option. Now, businesses of all sizes and in every industry are moving, albeit somewhat tenuously, to IaaS for their development and testing, production, backup and disaster recovery, employee- and customer-facing Web application (app) hosting, website hosting, high-performance computing, and big data analysis.
The best part about IaaS is that this instant, elastic IT environment enables organizations to quickly—and economically—scale up and down for temporary, experimental, or shifting workloads as needed.
Generally, IaaS customers pay on a per-use basis, typically by the hour, week, or month, though new options are emerging. Some providers also charge per amount of virtual machine (VM) space used, others require a predetermined usage reservoir, while still others package their hosting services with software and support in an effort to differentiate their products.
Regardless of the providers’ sales model, the underlying cost savings for customers remains one of the most compelling reasons for adoption.
With IaaS, “there’s no need to maintain and upgrade software and hardware or troubleshoot equipment problems,” notes a Microsoft spokesperson. “IaaS sidesteps the upfront expense of setting up and managing an on-site data center, making it an economical option for startups and businesses testing new ideas.”
Miles Ward, head of global solutions, Google Cloud Platform, adds that “organizations are able to trade upfront capital expense for variable expense while gaining massive economies of scale.” Now, a global app launch can take place in a matter of minutes, he adds. If it’s a success, the servers keep humming. If it’s a less-than-stellar kickoff, the number of servers can be instantly cut down to fit the traffic. With IaaS, there should never be a VM sitting idle, as a server might be doing in a physical data center.
John Engates, CTO, Rackspace, uses the “pets versus cattle” analogy to put this practice into perspective, “pets are animals you name; you keep them over their lifespan and consider them part of your family,” he says. “That’s how IT feels about their servers. They might even have names. But cattle serve a more utilitarian purpose. You don’t get attached to them—just like VMs.”
Temporary use cases of IaaS can be seen every day and they’re impressive. A Hollywood studio creates its special-effects blockbuster, like The Walk, on Google Cloud; the Olympics are live-streamed via Microsoft Azure; a presidential campaign spins up 200 unique applications on AWS. Then, once the need is over, these “emergency” data centers in the cloud are no longer part of the infrastructure. No one has to bother with them, much less pay to keep them running.
Many use cases are evolving for IaaS, too—like running a customer relationship management (CRM) system on CenturyLink so it’s accessible by sales staff 24/7, or maintaining stock-market-analysis services that use AWS’ massive size and power to reconstruct market events down to the individual record and analyze three-billion data points in 2.8 seconds.
The allure of acquiring and using infrastructure on demand and paying only for how much is consumed is a key driver of IaaS adoption. But there are side benefits as well, including improved business continuity, more stability, and better security, says the spokesperson for Microsoft. “With IaaS, the IT team is freed up to focus on its core business rather than infrastructure,” they say, echoing the sentiment of most IaaS product managers.
The IaaS Cloudscape
It’s true—security is now seen as a value-add for IaaS adoption. For the first time since cloud computing launched, security no longer ranks as the number one detractor, according to RightScale’s State of the Cloud. The report cites increased maturity of both users and providers for this newfound peace of mind.
“I think security really has been more of a perception problem than anything,” says Engates. “There’s the old mindset that if you give up control of your data, you’re putting it at risk. But the analogy I use is to talk about the difference between driving your own car or riding in an airplane. Statistically, there are far less airplane accidents than car accidents. In terms of IaaS, when you hand your data center over to somebody else who’s world-class at it, and has the resources and expertise to do it right, 24-7/365, your data’s security is going to be much greater in the hands of those experts.”
“Basically,” he adds, “your company’s not in the business of running a data center. IaaS providers are.” And, over the years, they’ve amped up their focus on security as the number of connection points to corporate infrastructures has grown and potential threats have increased. Today, customers find comprehensive security with built-in protection and monitoring from the chip through the stack, close adherence to government regulations, high-speed encryption, and solutions that provide support for unique networks to segregate public, private, and management traffic.
In its first iteration, open-source public cloud was great for collaboration and ad-hoc projects, but didn’t offer much security. Private cloud evolved as an attractive option for security-minded enterprises, especially in the financial, health, and other highly regulated verticals, but that meant having to buy and maintain software and infrastructure—something that IaaS intended to avoid. In response, some cloud service providers (CSPs) began offering private versions of their public cloud—and vice versa—blurring lines, raising confusion, and eventually ushering in “hybrid cloud.”
Now, by spreading the business over both public and private platforms in a hybrid configuration, an organization can easily interact publicly with its customers while keeping private data safe. The challenge of hybrid-cloud IaaS, however, is in its complexity and ensuring that all platforms—cloud-based and even physical data centers—interact efficiently.
Because of this, RightScale’s 2016 report notes that “lack of expertise” has supplanted security as the biggest pain point in IaaS adoption. As organizations place workloads in the cloud, and the average number of diverse clouds used by a single organization has risen to six, “managed cloud” has become part of the IaaS ecosystem.
Whether planned or created by rogue teams and third-party vendors within an organization, “the proliferation of cloud platforms requires pretty stringent oversight,” explains Engates. That’s why he says his company chose to stop competing with giant providers like Amazon or Microsoft in 2014, and returned to its roots—focused on its trademark “fanatical support.” It stopped selling its high-performing private cloud as a commodity and began bundling it with service and support.
“Managed cloud is in our DNA,” he says. “We are now focused on providing the service, management, expertise, and operational support that organizations need in all cloud contexts—but just don’t have the time to hire and train the talent. We help our customers on whatever platform they choose to be.” To the point, he notes, where his team supports competing CSPs.
Similarly, CenturyLink says it differentiates itself in this market with its end-to-end portfolio of products and services.
Partnering with Rackspace, CenturyLink, or any managed service provider, may not be the cheapest choice, Engates concurs, but it may be more cost effective in the long run, as mistakes—and cost overruns—are avoided. Management consultancies, like the “Big 4” accounting agencies, are moving to fill this void, too.
The Cost of (Non)Ownership
In an effort to stay competitive, pricing wars regularly break out among IaaS providers. If Amazon charges “x” per hour, Microsoft will meet or beat it, and the rest of the industry falls in line. But, as Engates has observed, this “race to the bottom” is causing some negative fallout.
The initial promise of ease, or turnkey, IaaS migration has left too many organizations dissatisfied with their initial decisions to move to the cloud. A recent survey from THINKstrategies and INetU showed that 43 percent of the respondents’ implementation efforts had either stalled or failed completely. Among those companies, nearly half had cost overruns, and most had to change their cloud environment within six months to a year.
CenturyLink’s Seroter knows this all too well. “If you think you’re moving because of money, you’re going to be disappointed,” he says, “because the total cost of ownership could be higher on a per-consumption basis.” But he hastens to remind his customers that, regardless of the cost, the business will likely gain in areas that might not be readily quantified, like entering new markets, keeping customers happy, scaling workloads, and doing things that couldn’t be done when IT was busy “racking and stacking” servers.
“It all comes down to ‘where do I invest my dollars and time?’” adds Seroter. “‘Do I invest ten percent more a year for compute—to get 50 percent more compute than I had before without hiring a single new person?’ That’s pretty compelling stuff.”
Gartner and other industry observers say cost management is highly recommended in order to achieve the best return on investment. They caution IaaS adopters to be on the lookout for unauthorized charges—and to make sure they’re using what they’re paying for. Too many companies forget to shut off the fire hose once the emergency is over—and continue to pay for unused servers. Keeping an eye toward lower-cost clouds is also recommended.
Suggested Migration Patterns
When consulting customers about migrating to IaaS, Engates says it’s important to consider the business priorities, the quick wins, and the potential problems. He’s also observed that, because large enterprises have greater resistance to change and more complex regulatory constraints, they tend to focus only on migrating tactical workloads such as test and development, where the degree of risk is perceived to be low.
“You have to get a sense of how aggressively the organization wants to move,” he says. “Is there an event—like a data center lease expiring—that’s behind the decision to migrate? Or are they just being proactive in recognizing this is a trend they have to jump on.”
He recommends that IT first inventory its applications, “and where they physically need to sit. Do they belong in the cloud, or in your data center? Does an app need to be rewritten to fit on a cloud server?”
That’s right—rewritten. In today’s run for the clouds, some key details have been omitted from in the marketing hyperbole. It turns out that applications may need to be modified to work properly in a cloud environment where the architecture is quite different from the one on premises, points out Engates. “In the cloud, you don’t get to select the equipment, hardware profiles, and storage capacity,” he says. “Instead, you have to rely on a pre-set menu, so some apps may have to be rewritten.”
Migration is certainly not just a “flip of the Frankenstein lever,” quips Google’s Ward. He acknowledges that migration takes time, and planning, and is better looked at as a sequential process that enables IT teams to learn as they go. “Google is straightforward and easy to use,” he says. Still, he recommends starting with the “low-hanging fruit”—something that migrates with the least amount of trouble and disruption to the organization. “You shift and learn,” he adds. “The more you do it, the aperture opens a little bit wider.”
It’s easy to acquire a simple IaaS service, industry experts agree, but properly calibrating the solution to support specific applications and workloads takes specialized skills that most organizations lack.
The big cloud providers are moving to fill this void. Offering support for enterprise-level migrations, Amazon released AWS Database Migration Service in mid-March, and Microsoft has announced that its database program, SQL Server 2016, will include technology for smoother transfers to Azure. In both cases, however, they expect customers to convert databases to their proprietary software.
For more customized migrations, it’s still probably best to hire a cloud consultancy, systems integrator, managed service provider, or other trusted advisor, notes Engates.
What’s the Downside? Downtime.
By choice or happenstance, most organizations spread their workloads over a number of different clouds via several cloud providers. Whether intentional or not, redundancy affords better security as well as a level of protection against outages. And outages do happen.
To battle this concern, Microsoft Azure now boasts data centers in 19 unique regions around the world, while Amazon, true to its name, will have the broadest reach by the end of this year. As 2016 nets out, AWS plans to have at least 40 “Availability Zones” within about 15 geographic regions around the world. Availability Zones are one or more discrete data centers—each with its own redundant power, networking, and connectivity—housed in separate facilities. That means each region has multiple Availability Zones and data centers establishing ultimate redundancy and ensuring fault tolerance, low latency, and a virtually fail-proof infrastructure.
As Seroter notes, “being single-sourced in the cloud is a risk, but sometimes the risk is worth it,” he recalls a major outage at Amazon that hardly registered for power-user Netflix, because it is purposely architected across numerous Availability Zones to virtually eliminate the risk of downtime. Social media outlet Reddit, on the other hand, was kicked offline for a few hours in that same outage, but the loss of business didn’t cost the company as much as paying for all those redundancies.
“In the cloud, you get the level of resilience that you want to pay for,” says Seroter, “This really all boils down to good architecture practices by the customer as well as honest conversations with the provider to find out where potential failures might lurk, and what’s needed to mitigate them.”
How High is the Cloud?
Mainframes still run many of the world’s largest companies, which probably won’t be 100 percent cloud-based for decades. Yet even among them, adoption of IaaS continues to grow in areas like disaster recovery, archival storage, and software development.
As with any disruptive innovation, it takes time for technology and corporate acceptance to simultaneously mature to the point of total acceptance. And, despite the hype, industry watchers unanimously believe that over time, very few companies will have data centers of their own. Enterprise-class IaaS technology will simply become the norm—as obvious as the electricity used to run it.
“We think of this as the construction of a net-new utility,” says Ward. “Real disruption in technology doesn’t happen until there is a native execution of that technology.” Keeping with his utility analogy, he points out that even though the use of electricity to light a city may have disrupted the gaslight market, it wasn’t until other, electricity-native inventions happened that electric power actually became disruptive. “It brought us the television, the phone—things that are radically different and have nothing to do with the original use cases. That’s what we’re seeing from some of our most cutting-edge users today. Truly, the sky’s the limit,” he adds. SW
Apr2016, Software Magazine