As workloads migrate back from public clouds, hybrid cloud grows

Cloud computing gained popularity for its ease-of-deployment and flexible resource consumption, and while that works for many critical applications, it is not a panacea for every app an enterprise supports. That’s why as cloud adoption continues to rise, some companies are opting to bring workloads back on-premise in certain scenarios.

Eighty percent of respondents to a recent IDC survey report repatriating workloads from public cloud environments. On average, organizations expect to move 50% of their public cloud applications to hosted private or on-premises locations over the next two years, IDC says. The top repatriation drivers, according to the research firm, are security (cited by 19% of respondents), performance (14%), cost (12%), control (12%), and to centralize or reduce shadow IT (11%).

The move from cloud to on-premise isn’t a precursor to reduced use of cloud, however. Rather, it’s evidence of the maturing market and the experience of cloud users.

Now that cloud has been an option for computing for many years, IT departments are seeing its benefits in the right use cases—and realizing when it’s not the right fit. Computing options are not limited to cloud or not: IT organizations can do both, and in 2019, the technology and tools to enable advanced management and orchestration between public and private clouds are more readily available.

“Hybrid is the new normal. Every major public cloud vendor has acknowledged that not all workloads are going to work in the clouds,” says Ajay Patel, senior vice president and general manager in VMware’s cloud services division. “Customers are going to run workloads where it makes more sense. It is less about public and private; now the pragmatic experience has kicked in. It is not all or nothing.”

Where to cloud? Think application first

Today’s enterprise IT teams are considering the best use of cloud when it comes to their specific app needs. Several factors from budget spend to data privacy to mobile app accessibility and resource consumption will drive the decision of where to locate an application when considering cloud as a compute option. And all those factors also contribute to the growing adoption of hybrid cloud, which is a mix of on-premises, private cloud, and third-party, public cloud services with orchestration between the two platforms.

“Enterprises have started taking an ‘application-first’ approach to their cloud deployments and plans,” according to the Enterprise Cloud Index, research conducted by Vanson Bourne on behalf of cloud vendor Nutanix. “Using an application-first approach, enterprises will set up their application to run, increasing, in the most economically appropriate location. That location will be based largely on the predictability of the workload.”

IT teams have been tasked to think mobile first, cloud first, and now application first. And it makes sense. Apps drive most businesses’ bottom line, and the strategy should be to support the apps to optimize performance. That means the market for cloud services is maturing, and cloud consumers are getting smarter about which apps will need on-premise support and which will benefit most from public cloud.

“Cloud makes more sense when you need to get the compute resources you need relatively quickly and cheaply, and for a large, established company, on-premise could make more sense,” says Domenic Alcaro, vice president of data center software solutions, Schneider Electric. “Security and bandwidth are often reasons to stay on-premise.”

The functionality of the app and the immediacy of fulfilling requests is a critical consideration for IT managers deciding where to best locate app resources.

“For traditional business apps that are fairly static, private cloud is starting to look more efficient. There is no reason to move those apps. And transactional, high-performance apps that run for the business, those are running in private cloud or being replatformed back,” says VMware’s Patel. “Apps that are targeted at specific initiatives, that are short-lived, bursting up and down—cloud is much more native in those scenarios.”

Cloud costs driving some repatriation moves

Cloud has the benefit of being considered a less expensive resource option than rolling out a fully resourced data center. But depending on the cloud deployment, adopters report they aren’t always spending less money or being more cost efficient when turning to cloud.

According to the Vanson Bourne research, just 6% of organizations that used public cloud services reported staying under or within budget, yet another 35% said they overspent, exceeding planned budgets. Not being able to predict the potential cost of a cloud deployment could also be driving a move back on-premise.

“Cost is a strong driver for both colocation and public cloud. This suggests that customers’ cost experience with either option is variable. Colocation’s more predictable costs per month is an advantage, driving 39% of customers out from a public cloud,” according to a report from 451 Research commissioned by Schneider Electric.

“You have to be very good at managing your cloud, aware of all the potential associated costs, and keeping control of it, or in some cases, you will continue to get charged after you finish a project,” Schneider Electric’s Alcaro says. “In some cases, it is better to keep it on premise so you can clearly understand, ‘here is what it would cost.’”

Vendors such as CloudPhysics, for instance, will perform assessments to help IT departments to better understand what resides in the cloud, the costs, and if it would make sense to move it back on-premise. But again, keep cost in mind: “I have seen situations in which it’s free to send data into the cloud, but if you want to bring it back on-premise, there is a fee per gigabyte to do that,” Alcaro says.

Cloud at the edge

The rise of edge computing, locating compute resources closer to where they are being used, could be another reason companies are opting out of cloud deployments.

For one, it could reduce the bandwidth costs that are increasing for many cloud users. Apps and processing happening on the cloud are consuming bandwidth that cloud adopters didn’t factor into the budget. With edge computing platforms, data aggregation and processing happens at the edge, meaning data isn’t passed to the cloud, just the results of the processing are sent to the cloud. That type of processing lowers bandwidth costs for companies by focusing that process on local computing platforms.

And edge computing could also better support emerging technology trends such as the Internet of Things and the myriad new devices connecting back to enterprise IT departments. Take, for instance, the trend toward smart cars, or automobiles loaded with intelligent software for directions and other functions.

“For instance, if the app is on a car that needs to make a decision on whether to turn left to avoid an accident, then the technology has to be on the car or very close to the car, it can’t be in the cloud. It would be too late if the decision was made in the cloud,” Alcaro says.

Also, companies that support branch offices could depend on edge computing to enable enterprise-level resources in those remote areas that don’t support a full-blown data center.

In the big picture, cloud adoption is still accelerating, and the sophistication of cloud applications and services is still increasing, IDC asserts. At the same time, enterprises continue to increase their investments in on- and off-premises private cloud solutions to improve security, performance, cost, and control requirements.

“The modern enterprise wants to extend the value of its classic applications. It has multiple options as it evaluates best fit and as its technical expertise with emerging cloud technologies improves,” IDC reports. “All enterprises have more opportunities and capabilities to migrate or extend applications by leveraging alternate landing zones.”