When cloud repatriation makes sense
Even the cloud has its limits. In certain situations, a traditional data centre is still the best place to host one or more applications
15 April 2019 | 0
A growing number of enterprises are pulling selected applications out of the cloud and returning them to their brick-and-mortar data centres. Cloud repatriation is gaining momentum as enterprises realise the cloud isn’t always the best solution to IT cost, performance and other concerns.
Dave Cope, senior director of market development for Cisco’s CloudCenter, believes that technology has evolved to the point where enterprises now have the unprecedented freedom to locate applications wherever maximum cost, performance and security benefits can be achieved. “There’s an ability to place workloads where they best reside based on business priorities, not IT constraints,” he notes. “We’re starting to get this natural distribution of workloads across existing and new environments… where they make the most sense.”
If your organisation’s cloud infrastructure is beginning to feel more like a cage than a boundless resource environment, consider these instances when returning selected applications to a traditional data centre may make sense.
1. To save money
Many organisations can achieve substantial savings by reducing or eliminating the high recurring operational expenses of the public cloud, observes Jeremy Kurth, CTO of US IT services provider Winxnet. “A reduction in total cost of ownership is one of the potential benefits in repatriation,” he says. “Public cloud offerings can provide value-add compared to on-premise solutions, but it typically comes at a premium in recurring expenses.”
Organisations that felt the public cloud was the answer to all their problems are now often finding that the expense and advantages just aren’t as relevant or applicable to their specific situations, Kurth explains. “Considering the added true all-in cost over a two- to three-year period, the compelling argument to stay with public cloud just doesn’t add up over time, especially when compared to cheaper alternatives now available in the market.”
Cost is generally the primary reason for shifting cloud applications back to data centres, observes Chris Carreiro, CTO of data centre monitoring and support firm Park Place Technologies’ ParkView platform. “If the assumed cost is greater to operate [in the cloud], it might make sense for a company to bring back an application or service to on-premises or colocation,” he says. “Cost of memory and disk [resources] continue to decline, changing the requirement to operate in cloud, allowing a company to be more flexible.”
Travis Morrison, director of IT at craft brewer New Belgium Brewing, which recently migrated its core applications from an off-premises managed cloud to on-premises Dell EMC PowerEdge servers, says that reduced cost and maintenance expenses played a major role in bringing the software back on site. “We wanted predictable costs when scaling, and we have a talented staff that can manage on-premises equipment,” he explains. “Additionally, ROI for cloud diminishes with a hyperconverged stack, as maintenance is simplified.”
As the cost of using public cloud services continue to grow, improvements in both performance and budgeting can be achieved by repatriating selected applications. “By leveraging available and cost-effective solutions, such as [Windows Server] Storage Spaces Direct, hyperconvergence, storage-class memory and software defined networks (SDN), the gap in flexibility, scalability and redundancy that once made public cloud offerings so attractive has greatly diminished,” Kurth explains.
Applications that need to access large amounts of data, such as data mining tools, can also benefit from a move back home. “Some applications move data in and out in large amounts frequently,” Cope notes. “This can be costly with most [types of] cloud billing.”
2. To gain more control over applications
Prime candidates for repatriation are heavily used applications offering primarily static functionality. Such software can often be maintained more effectively on-premises, where its environment is a controlled state and costs are fixed and predictable, says Vinod Pisharody, CTO at network hardware firm Array Networks. “Repatriation provides better control over the applications and enables IT to [better] plan for potential problems,” he explains.
The prospect of stronger control was a major reason why New Belgium Brewing brought its core applications in from the cloud. “We were able to better control ERP performance, monitoring and troubleshooting,” Morrison says. “They are our enterprise applications and are a good fit for co-location and hyperconvergence.”
Additionally, in the event a cloud provider’s promised flexibility and management benefits fail to materialise, repatriation can make solid financial sense over the long term. “The cost of staying in the cloud cannot be justified and the application should be moved on-premises where the environment can be tailored to suit it,” Pisharody recommends.
3. To enhance application performance
Repeatedly failing to meet critical operational benchmarks is a sign that a cloud-based application might perform better on-premises. “Applications that are latency sensitive, have long-running I/O intensive periods or have datasets that are large and require transport between various locations for processing are generally prime candidates for repatriation,” advises Jeff Slapp, vice president of cloud and managed services at 365 Data Centres.
Mature applications are also frequently prime candidates for repatriation. It’s common for enterprises to turn to the cloud to evaluate new business tools and proof-of-concept projects. Yet after the review period ends, the application may find a more suitable home back in the data centre, Carreiro notes. “Once they grow and stabilise, they are [often] seen back in the data centre,” he says. “Predictable applications and business mission critical applications generally will find their way on-premises if IT/IS is planning a shift.”
Applications or workloads that depend on very high level of performance, integration, customisation or larger datasets can often be prime candidates for repatriation. “In general, the less ‘normalised’ or ‘standardised’ a workload is, the less attractive a public cloud offering becomes,” Kurth notes. This is because a public cloud is primarily built for scale, which typically requires a strict policy of standardisation. “Most public cloud providers have to achieve scale in order to justify their investments and they do this by catering to the average setup,” he explains. “When a solution or workload does not adhere to the industry average, the added costs and reduced benefits of the public cloud [needed] to accommodate the ’round peg into the square hole’ situation are often unattractive.”
4. To achieve stronger security
Building an attack-proof cloud security architecture, particularly within a complex multi-cloud environment, can be both challenging and costly. “Cloud repatriation may offer more secure environments and the chance to address multi-cloud issues,” says Carl Freeman, executive director of the cloud and digital advisory at business consulting firm EY.
Kurth believes that on-site infrastructure can be isolated and designed to present a smaller surface area for targeted attacks than a typical large public cloud. “When done correctly, this design can lead to improved ‘security through obscurity’ results,” he notes, referring to the controversial notion that the fewer people that are aware of an application and its security measures, the less likely it will ever appear in an attacker’s crosshairs.
5. Simplified regulatory compliance
In an IT world that’s becoming increasingly regulated by various international government bodies, each with their own rulesets, running applications on premises at a specific site can make life easier and simpler while lessening the possibility of accidental non-compliance. “Let’s say an application went to the cloud, and then GDPR compliance requires changes to data retention or use—repatriation would benefit that app,” Cope says.
Cope believes that positioning location-sensitive applications at a defined place is one example of what will eventually become a natural distribution of workloads across public and private clouds and on-premises data centres. “So repatriation could become a part of both industry and technology maturity,” he explains.
Proceed with caution when moving off cloud
While repatriation can be tempting, it’s also a radical move with serious financial and operational implications. “Cloud vendors generally make it very hard to move off the cloud both terms of the cost and contract,” Pisharody says. “In addition, initial expenses for setting up an on-premises environment if it has been dismantled, or if one never existed, can be high.”
Organisations dragging applications back home may also encounter numerous technical challenges. “By migrating applications back to the on-premises data centre, organisations give up scalability, flexibility, availability and elasticity from the cloud,” Freeman observes. “Additionally, on-premises data centres may not actually be more secure than the cloud, given evolving internal threats.”
Then there’s the not-so-simple matter of unravelling all the interconnections that currently exist between applications, data and the cloud itself. “In general, the repatriation is going to be painful,” Pisharody warns. “Separating the portions of applications that have dependencies on the running environment versus the portions that simply manage data can likely help minimise disruption,” he says. “If you have architected into the cloud specific services, there will be a certain amount of rearchitecting to be able to move the workload back to a data centre,” Cope adds.
Additionally, enterprises that rely on big data for tasks ranging from predictive analytics to security analyses may find themselves facing a mammoth moving job. “Moving all that data is a tedious process,” Cope observes. Such organisations may face the choice of leaving their data in the cloud or beefing up their on-site storage and storage backup capabilities.
Careful planning, thorough research and staff training and preparation are generally regarded as necessary for a smooth repatriation process. “Know your staff’s technical capabilities and understand your business and whether cloud or on-premises is right for you,” Morrison advises.
“In most cases, cloud repatriation should be a last case scenario,” Freeman cautions. “Organisations should consider all cloud solutions, such as multi-cloud, before returning [applications] to on-premises data centres.”
IDG News Service