ra2 studio - Fotolia
As they find that a mix of cloud providers can meet their needs better than one, some enterprises pursue multi-cloud computing. But while this approach offers promise, it can also cause plenty of IT headaches.
Portability is one of several multi-cloud challenges that negatively affects adoption, said Lydia Leong, analyst at Gartner. Each cloud platform has its own strengths, but it can be cumbersome to attempt a lift and shift from one cloud provider to another.
"If you are running a big digital business in AWS and scaling to millions of users and you also have productivity apps in Azure, it will be two very different styles of computing," Leong said.
Multi-cloud users are largely divided into two segments, said Deepak Mohan, analyst at IDC. Small companies often choose multi-cloud computing to access the cheapest or best new service to meet a particular need. For example, a smaller company might shop around and find the lowest-cost instance type on Google, he said. Larger organizations typically adopt a multi-cloud strategy because they don't want to be tied down to a particular provider or exposed to risk.
"A big bank might decide they want to have a [multi-cloud] backup plan, especially given the favorable pricing of cloud, which is far below what they would have spent in the past on physical backup sites," Mohan said.
An organization, for example, could host its primary operations on AWS and choose another cloud provider as a backup platform -- but that approach can make it harder to provision, monitor and manage resources. Problems can especially arise when you use AWS-native management tools, such as CloudWatch, which don't work on other cloud platforms.
Multi-cloud skills are hard to gain
Jason McDonald, president at Contino, a global technical consultancy, said that multi-cloud computing can pose a significant challenge for users, especially for those without a strong cloud architecture design and automation strategy.
"The skills and knowledge required to take full advantage of the AWS platform requires experience, experimentation and time," he said. And that's just for one cloud; multi-cloud expertise is considerably more difficult to achieve.
"I've never come across a customer that was able to operate at maximum efficiency across multiple clouds due to the massive pace of innovation and focus required to continually improve and enhance the operating model," McDonald added.
Third-party cloud management tools can help with cross-platform support. Tools such as HashiCorp Terraform or Serverless become especially important as users move to more sophisticated services, like AWS Lambda, Azure Functions or Google Cloud Functions, and require an abstracted management layer. Containers also make it easier to encapsulate a full environment that can spin up across multiple clouds, said Eric Johnson, AWS evangelist at Rackspace.
A simple approach is the best approach, according to McDonald. Don't try to immediately incorporate serverless, microservices or machine learning capabilities in multi-cloud computing; the learning curve with those technologies will slow adoption and business-related outcomes.
"Build these capabilities over time, and then [add] more advanced services and automation capabilities to eliminate mundane tasks that are not differentiating to the business," he said.