When looking through a selection of cloud native technology, whether it’s vanilla Kubernetes, a closed-source software for cloud native apps or an obscure open source project, chances are that the ability to operate across clouds is listed as one of the benefits of using the technology.
Given how pervasive the multicloud message is in the cloud native ecosystem, it would be logical to think that operating in multiple clouds is an uncontroversial, standard best practice in the industry. Perhaps you wouldn’t find the major cloud providers advocating for multicloud, but everyone else would think it’s unambiguously a great idea.
Yet this isn’t actually the case. Sure, there are many people — and companies — who maintain that running in multiple clouds is the best way to reach as many customers as possible and mitigate financial risk related to lock-in. But not everyone agrees.
Pros of Multicloud
“Multicloud is almost being the default cloud strategy as organizations look to avoid vendor lock-in,” explained Nicholas Dimotakis, vice president of field engineering, datacentre at Canonical. In fact, avoiding lock-in is a common rationale for running in multiple clouds. The implication is that running in multiple clouds will be cheaper than running in just one cloud because you’ll have better-negotiating power with each cloud provider and better ability to price shop. “This is clearly important for enterprises, as cost optimization was seen as the biggest driver for adopting a multicloud setup,” Dimotakis said.
That’s not the only reason, though. Cloud services are not 100% commoditized, so each public cloud has different infrastructure and capabilities in their ecosystems. “Instead of being restricted to the ecosystem of one vendor, a multicloud approach permits organizations to deploy a mixture of cloud apps to suit their needs across the business,” Dimotakis said. “The beauty of multicloud is that it provides the flexibility to spin up whatever resources are needed to solve specific needs.”
Cons of Multicloud
According to Erik Peterson, founder and chief technology officer at cloud cost management services provider CloudZero, technology decisions should come down to three factors: Is this going to help my team move faster? Is it going to ease or add to my operational burden over time? And what is going to be the most cost-effective over time? “If you’re a technology organization, you should never find yourself making a multicloud decision because multicloud runs afoul of all three of those,” he said.
The problem, according to Corey Quinn, chief cloud economist at the Duckbill Group, is that some people think that they will save money by running in the cloud. “Let me save you some time,” he said. “No one saves money in the cloud. It’s a capability story.”
And therein lies many of the problems with multicloud. To make workloads portable between clouds, they can’t take advantage of any of the cloud provider’s proprietary capabilities. Those capabilities are where the magic happens, so dumbing down the workloads so much generally means that organizations can’t take advantage of the cloud’s capability advantages, Quinn said.
So When Should You Use Multicloud?
Quinn, is not generally a fan of multicloud but concedes that there are situations in which it makes sense to operate in multiple clouds — though under very specific circumstances. One is if a company acquires another company that uses a different public cloud — migrating to the parent company’s preferred cloud provider shouldn’t be a major priority.
“You could also make an argument for specific workloads belonging on different providers if they’re not tightly coupled to other workloads,” he said. But, he cautions, it’s rarely a good idea to build something from scratch with the idea that it will be able to seamlessly run anywhere.
What about situations where your customers insist on you not being in a certain cloud? “If you want to sell to Walmart, you can’t use Amazon Web Services. I know that’s probably true,” said Peterson, referring to how the two companies are direct competitors in the retail space. “But that’s not really about Walmart or AWS, it’s because you work in a business where technology choices are political. That’s not an argument for multicloud, it’s an argument for no cloud. If in your business technology choices are political and being on the wrong cloud could be bad for your business, you shouldn’t be on the cloud at all.”
In other words, those companies should really just stay in the data center, or at least keep some of their capabilities in the data center. “I think running your own on-premises data center is going to be more efficient and less expensive than taking a multicloud approach,” Peterson said.
Perhaps just as key to this idea, though, is that the ‘Walmart problem’ is something that should be dealt with as it arises. Don’t plan your cloud strategy based on the potential of selling to Walmart at some point in the future.
Do You Fear the Cloud Providers?
According to Quinn, the reason so organizations are so interested in multicloud comes down to fear. Instead of seeing cloud providers as true partners, they see them as partners of convenience who might turn on them at any moment, increasing prices dramatically, discontinuing services you rely on or otherwise making a unilateral decision that ruins your business. Or there’s a lack of trust that the cloud provider will provide the services it promises to, leaving you stuck.
“I’m not going to say that there’s not a valid psychological story, but the arguments against the public cloud are very different than they were 10 years ago,” he said. “Maybe you don’t trust the public cloud. But your tax authority does, your bank does and your regulator does. What makes your workload more special than any of those workloads?”
Again, if you fundamentally fear the public cloud, it’s possible that’s more of an argument for an on-premises data center than it is for a multicloud approach. If the public cloud scares you, why would it be better to work with several of them at once instead of working with none?
Perhaps the multicloud fever is a reminder, though, that not all technology decisions are made rationally, even when we’re talking about business to business enterprise decisions.
Amazon Web Services is a sponsor of The New Stack.