All signs point to a cloud computing future where nearly all software, data, and resources are accessed over the Internet. But we are far from there yet. While many companies take advantage of some cloud-based software solutions, many still have their own information technology infrastructure for storing data, running operations-critical applications, and connecting far-flung offices, among other tasks. They may also be wary of moving computing resources to the cloud, fearful of the costs of a migration and losing control of what they may see as a competitive advantage.
But as cloud solutions and their ecosystems mature, it becomes harder for organizations to cling to the old IT model. Why are companies adopting public cloud solutions, essentially buying a “slice” of a server in a cloud-computing environment that is shared with other clients?
Part of the equation is cost. For many companies, public cloud providers can supply IT infrastructure and applications services at a much lower cost than providers of on-premises solutions. That’s not just because of lower hardware costs, but also because of lower management costs, says Timothy Chou, a lecturer in cloud computing at Stanford University.
“The cost of the server is not really the cost of the box itself, but the management of the server—managing availability, managing security, and managing performance,” Chou says. “The cost to manage the box is probably four times the cost of the box per year.”
Public cloud providers also improve reliability by replacing much of the human labor with automated management. There is no clear dividing line designating when public cloud services are a better choice than company-owned data centers or on-premises IT, Chou says: For companies with fewer than 1,000 servers, the obvious choice is the public cloud, but companies with more than 100,000 servers (for perspective, Google has about 1 million servers) will obviously want to handle their own hardware.
The growth of applications in the public cloud comes from two sources, says Deepak Mohan, a research director at International Data Corp.: 1) new applications that are born in the cloud and 2) applications that are moved, typically when their on-premises hardware nears the end of its life or when it’s time to scale up.
The motivation for moving an application to the cloud isn’t always cost, Mohan says. For enterprises that want to pilot applications, or for startups or small companies launching new applications, the risks are lower in the cloud, and organizations can avoid the upfront costs of leasing a data center, hiring people, and buying equipment.
But even mature companies that aren’t looking to avoid upfront costs are finding benefits from the “cloud ecosystem” — the value-added services at the infrastructure layer. A cloud platform does not just virtualize computing resources, storage, and networking; it also allocates those resources as necessary. Value-added services can include monitoring and auditing reports that can help a company decide whether to offload other, similar tasks to the cloud.
“Increasingly we see, especially with more-mature companies, that while price is important, the ecosystem and the other benefits that cloud services provide are growing in terms of importance,” Mohan says.
Business advantage, such as increased agility or faster delivery of products and services, is the primary reason that many companies move to the cloud, says Lydia Leong, a vice president at Gartner who covers cloud migration. When companies are planning a move to the cloud, they should also be planning business transformations to take advantage of automation.
“[Migrations] have to be carefully planned, and there is upmarket expense for that migration,” she says. “You’re making an investment to save money over the long term.”
But just putting existing IT systems in the cloud is really inefficient, Leong says. “The cloud enables you to drive a great deal more automation, but if you are just effectively shifting from buying to renting, that’s not necessarily to anyone’s advantage.”
The top two public cloud providers are Amazon Web Services (AWS) and Microsoft’s Azure, with a substantial drop-off to the next tier, Leong says. Rackspace is essentially out of the public cloud provider business, although it manages other companies’ cloud implementations and migrations. Google has been very interested in serving innovative companies, but less interested in providing a “your mess for less” public-cloud service broadly for customers. Oracle and IBM seem to be aiming for some niche markets, but “at least at this stage [are] not really poised to be broad competitors,” Leong says.
A majority of organizations migrating to the cloud are migrating to AWS because its ecosystem is more mature than Azure’s, Leong says. “It’s a lot easier to get help and find people who are experienced in doing that migration; it’s easier to license software and get support for commercial software that enterprises use on AWS.”
However, Azure has been improving in these areas. In posts on the G2 Crowd software review platform, Azure users say they like the ability to build and deploy servers at multiple locations across the U.S. at the click of a mouse and the ease of linking virtual networks. One user also notes that Azure allowed setup of Linux servers in its cloud last year, “which has made most arguments against Azure moot.”
Public-cloud pricing can be complicated, with variations based on what types of servers are offered, the size of the offering, performance, and geographic location. But the prices are publicly available, thanks to Amazon’s retail philosophy: It’s easier for customers to make the purchase if they know the price, and public pricing means no expensive sales force, says Stanford’s Chou. Microsoft has pledged to be price-competitive against public AWS prices.
How do organizations migrate to the public cloud? Companies that are migrating should enlist assistance from vendors that have experience in moving corporate IT to the cloud, says Leong.
“Most companies starting on this journey don’t know what they don’t know, and having experts from the very beginning tends to make projects more cost-efficient over the long run,” Leong says.
Most companies aren’t facing an all-or-nothing shift to the cloud. Some want to take advantage of infrastructure-as-a-service (IaaS) from cloud providers while balancing compliance and control requirements. Therefore, they may move just a fraction of their applications to the cloud, says Edward Wustenhoff, chief technology officer at Burstorm, which makes an app that models cloud deployment and infrastructure scenarios.
If a company opts for a cloud model, it also needs to decide whether it wants a public cloud model — shared, multi-tenant, with very little control — or variations of more control offered through managed service providers like CenturyLink and Rackspace, Wustenhoff says. (See “What’s Available in the Cloud?” below.)
For certain highly specialized applications, companies can improve availability, security, performance, change management, and bottom-line economics by optimizing their computing infrastructure (instead of deploying it in the cloud). But “if I don’t need control, don’t care as long as it runs any flavor of Linux, and can run it online, then for $20 a month I can get a very high-performance machine,” Wustenhoff says.
The decision-making doesn’t stop once an application is migrated. Having deployed an application successfully, a company will have to continually re-evaluate whether it should fine-tune its level of cloud services to fit the business model. “Somebody told me once it’s very cheap to fail in Amazon, but it’s very expensive to succeed in Amazon,” he says. “If you fail it’s okay; you turn it off and you don’t pay anything. But if you’re successful in that model, then you find yourself paying a premium for a lot of services because once you are successful, you start consuming more and more and more [resources].”
Before embarking on a migration project, of course, CFOs should have a clear understanding of the motives and capabilities of their IT departments, Wustenhoff says. “We see a lot of situations where the internal IT department … doesn’t understand or is not familiar with the capabilities that exist in the cloud.”
CFOs also need to have an open mind about potential cloud solutions, and periodically take the pulse of what’s happening in the market. Cloud technologies and business models based on those technologies are changing so rapidly that companies should be checking up on a quarterly basis, Wustenhoff says.
“We have examples where companies could cut their costs in half if they would move to a different [computing] model,” he says. “They don’t realize that and they say: ‘Well, there’s still a lot of work; there’s a lot of effort to move.’ I totally agree with that; it’s not trivial. But if it is truly a 50% cost reduction, then it might be worth it to go down that path,” he says.
As migration to cloud platforms increases in the next few years, CFOs may find the process getting more complex. That’s because the easy-to-move workloads are already in the cloud for many companies—in essence, the low-hanging fruit has been picked. In the coming years, in contrast, if and when they get comfortable with the public cloud, CFOs will be weighing in on migration decisions for mission-critical processes and new business ventures, says Allan Krans, the cloud, software, and data center practice manager for Technology Business Research Inc. (TBRI).
However, transferring mission-critical applications to the cloud will focus a lot more attention on cloud platforms’ security and performance.
In February, Amazon suffered an interruption to its S3 cloud data storage services, and a cascading effect knocked out several AWS services and a large chunk of the Internet. The outage took several large websites offline and affected other websites — including Netflix, Reddit, Adobe, and the Associated Press — over an 11-hour period. The cause of the outage: human error. An S3 team member entered an improper command, removing a larger number of servers from service than was intended.
Following the S3 outage, a survey of decision-making executives by TBRI found that, since moving to the public cloud, only 6% of the respondents had experienced more outages than they had expected, causing them to consider alternatives to the cloud. About 18% of respondents had experienced no public cloud outages, and 48% experienced fewer outages than they had expected.
The survey results indicated that many companies seem to have built-in expectations of public cloud outages that are in line with reality, Krans says.
When weighing the security risks of public cloud solutions versus company-owned IT, companies should consider country residency requirements for storing data, Health Insurance Portability and Accountability Act requirements, and payment card data requirements, Mohan says. Public cloud providers have built tools and constructs addressing these requirements, and companies that use the public cloud need to familiarize themselves with the tools so they can build the same level of security into their application layer.
Cloud providers can spend more on security measures than most of their customers can for their own IT, and most enterprises find that cloud security protocols are on par or better than their own, which lessens security concerns, TBRI’s Krans says.
To help decide whether a cloud provider is secure enough, customers may want to define their requirements and ask if the service meets them, Chou says. For example: “Within 24 hours of the release of any security patch, it goes through 1,000 tests and then is put into production within 39 minutes.” Or, “from the time an employee is terminated, his or her access to the building and the servers is revoked within 42 minutes.”
Those are two made-up examples, Chou says — no one is currently asking for those specific requirements—but they could help define just what “secure” means for the cloud customer.
“It’s the same as saying: ‘I have a fast car,’ and you say: ‘Well, that’s cool,’ or I say: ‘My car goes 0 to 60 mph in 2.5 seconds and it has 953 horsepower,’ and you say: ‘Wow, you do have a fast car,’” Chou says. “Today, people say: ‘Well, I have a secure system.’”
One reason some companies resist cloud adoption is the social factor, Chou says. Companies are wondering, “Who are these people, and if something happens, whom do I call?” he says.
Chou sees company-owned data centers eventually going the way of corporate remote networks that companies once built and managed themselves. “Today, there is no network engineer’s throat to choke when something goes wrong,” Chou says. “The flip side is that companies get new features and far better pricing than they would if they tried to do everything on their own.”
As more companies move to the public cloud in some form, social obstacles will matter less and less. The cloud is not just an IT cost-savings tool, but a better computing platform for organizations that want the ability to grow, to diversify, and to adapt rapidly to competitive threats and opportunities.
Keith Button is a freelance writer based in Valley Cottage, New York.
Infrastructure as a Service (IaaS)
Contains the basic building blocks for cloud IT and typically provides access to networking features, computers (virtual or on dedicated hardware), and data storage space. Provides the highest level of flexibility and management control over IT resources.
Platform as a Service (PaaS)
Removes the need for organizations to manage the underlying infrastructure (usually hardware and operating systems) and allows them to focus on the deployment and management of applications. Resource procurement, capacity planning, software maintenance, and patching are all taken care of.
Software as a Service (SaaS)
Provides a completed product that is run and managed by the service provider. A common example of a SaaS application is web-based email, where the organization does not have to manage feature additions to the email product or maintain the servers and operating systems.
Source: Amazon Web Services
Analysts project rapid growth for cloud computing offerings.
While the arrow for cloud adoption is pointing up, estimates for current public cloud use vary. Gartner pegs the figure at about 20% of all virtualized workloads, while Technology Business Research estimates that less than 10% of corporate IT spending is in the cloud. Oracle co-CEO Mark Hurd predicts that 80% of corporate data centers will disappear by 2025 as more and more companies shift to the cloud, freeing up their IT budgets for innovation.
International Data Corp. predicts that revenue for public cloud services worldwide will grow about 20% annually, reaching more than $195 billion in 2020. Of total public cloud revenue, 84% now comes from cloud software, including the components of software-as-a-service and platform-as-a-service, and 16% from infrastructure-as-a-service. According to IDC, manufacturing, banking, and professional services are the leading industries in cloud spending, at nearly one third of the total.
The top two providers of public cloud services, Amazon and Microsoft, recorded $12.2 billion and $14.4 billion in cloud revenue, respectively, in 2016. Morgan Stanley analysts predict those revenue figures will rise to $34.6 billion for Amazon and $46.6 billion for Microsoft by 2020, for compound annual growth rates of 30% and 34%, respectively. Morgan Stanley predicts similarly steep cloud revenue annual growth to 2020 for Google, 44%; Oracle and IBM, 33% each; SAP, 25%; and Alibaba, 80%. But none of those second-tier cloud companies are predicted to break $12 billion in cloud revenue by 2020. —K.B.
Are you moving all or most of your company’s IT systems to the cloud? James Eliason, CFO of Datawatch, a data analytics firm, has some advice about making the transition:
1. Ensure there is tolerance for cloud computing in the corporate culture. Skepticism and uncertainty among the C-suite is inevitable. We encountered some pushback with our proposed cloud initiative. But once the critics were able to see how the cloud would drive growth and value—and recognized that the initiative was being driven by the guy whose job it is to manage risk in the company—we were able to forge a consensus.
2. Make sure the user base is technologically capable. Cloud computing is supposed to make IT simpler. But any technological change requires some degree of retraining. A complete cloud transition may not be a good fit if employees are not technologically sophisticated.
3. Vet the cloud provider. It is critical to perform due diligence: check the vendor’s references; assess its financial health; evaluate its IT infrastructure for redundancy, uptime, and recovery; understand its onboarding processes; and clearly establish enforceable service-level agreements.
4. Make sure you have a reliable, robust “pipe” into your organization. Your cloud vendor’s uptime and reliability will be all for naught if your Internet connection goes down. Apply the same due diligence to your service provider as your cloud vendor, and make sure you have redundant, automatic failover connections.
5. Think globally. If your company competes globally, select a cloud vendor with international reach. Having to deal with multiple vendors serving different regions can introduce unnecessary complexities and uncertainties when problems do arise. You want one neck to choke, not several.