The move to cloud computing continues its steady growth, but sometimes an organization decides to retreat from the cloud or to move certain workloads back on premises. While not a widespread phenomenon, a cloud exit is a way for businesses to regain control over spending and better manage workloads they believe are struggling in a public cloud environment.
“As workloads grow in scale and complexity [in the cloud], they require more services to support,” says Hyoun Park, CEO and principal analyst at Amalgam Insights. This can quickly double and triple an organization’s costs — even for relatively predictable workloads. It’s no accident that AWS generates operating income in excess of $2 billion per quarter, Park said.
Companies consider a cloud exit strategy for various reasons. Sometimes on-premises or private cloud deployments make it easier to support a particular workload, such as one that is predictable but that has complex requirements, Park said. A cloud exit might also be of interest to a company that wants to choose its own hardware for specialized applications.
When to consider a cloud exit strategy
In fact, according to Forrester Research, the top six reasons for cloud adoption are:
“If you think of these drivers, infrastructure becomes incidental,” Chhabra said.
Even so, specific situations can lead a business to look carefully at a cloud exit strategy, said Alex Sokolov, senior software engineering manager at iTechArt Group, a custom software development company that has helped clients with both cloud migration and cloud repatriation.
A rollback might be needed, Sokolov said, when a large enterprise scales fast enough to justify the cost of on-premises data center management or when an implementation project requires hardware that complicates data transfer to the cloud.
Melanie PoseyAnalyst at 451 Research
In addition, organizations that have complex architectural maps sometimes miss dependencies that introduce latency and failure points in a cloud environment. This can happen to organizations of all sizes and in any industry, said Gartner analyst Brandon Medford.
Also, a change in regulatory exposure for an application can be a reason to explore repatriation, he said.
How to tackle a cloud exit
While each rollback is unique and requires customized steps, Sokolov sees five key points that should be factored into any cloud exit strategy.
- Plan it carefully. “Shifting back to an on-premises system is a long-term process. That’s why you need to draw up a thorough step-by-step plan and get ready to face a lot of challenges,” Sokolov said.
- Prepare to deploy. Automate deployment and conduct a complete test so you’re sure your application deploys rapidly and reliably.
- Ensure data transfer. Any migration means there will be some time when data will exist both in the cloud and on premises. “Such a situation usually causes some unnecessary delays in interconnection; you need to make sure that your app will be able to tackle them,” Sokolov said.
- Take charge. Get ready to deal with everything that was previously the responsibility of a cloud provider, such as infrastructure and security.
- Back it up. “You need to come up with a backup plan that will protect your business from such tiny but deadly accidents as a power cutoff.” A hybrid computing model could help, Sokolov said.
It’s also important to take time to identify snags that could snarl a cloud exit process, said Melanie Posey, an analyst at 451 Research, which has wrestled with how to describe the cloud repatriation trend.
“Anytime you are moving a workload anywhere, it just makes sense to ensure you have a business continuity plan so you can minimize or eliminate any downtime that might be associated with that move,” Posey said.
Whether the migration is to the cloud or from the cloud, make sure you map all dependencies. “You don’t want to have any unintentional downtime situations where the application breaks entirely,” Posey said.