Everyone is talking about hybrid cloud strategies, implying this framework is how all IT resources will be deployed in the future. It sounds great — on paper. Proponents tout hybrid cloud’s ability to tap into both public and private resources to build scalable applications that can be managed and secured as if they were being delivered privately — a way to have your cake and eat it, too.
Unfortunately, for the vast majority of workloads, this scenario will probably never play out. There may be a few isolated instances or use cases where this is workable, but, seriously, most workloads don’t need this level of flexibility, especially when stacked up against the considerable management and security challenges of a true hybrid cloud.
Analysts and pontificators — myself included — have talked about hybrid cloud strategies, but in reality, we are getting it wrong. Those who viewed hybrid cloud as a single entity — public and private — simply glossed over the business realities of creating, deploying and, most importantly, securing applications in a blended environment. Those who viewed hybrid cloud strategies as a mix of public clouds and private clouds — each with its own domain, where applications reside in one or the other — incorrectly labeled that architecture hybrid cloud. In reality, it’s multi-cloud.
Overlooking the challenges of hybrid cloud strategies
The first group, which believes a single cloud can span both public and private domains, tends to gloss over the different aspects of operating and managing that resource. How are administrators going to ensure security across multiple domains? What about the management tools that will vary between the two platforms?
Simply put, how many of us who switch between Mac and PC get frustrated by “delete” and “move to trash,” which are different ways to say the same thing? Now, imagine that at cloud scale. While many people like the idea of the elasticity of a hybrid cloud to scale out almost infinitely, how many applications truly need that much capacity on demand?
There are a few use cases where hybrid cloud strategies might make sense, primarily bursting applications out during peak load. The overly tired examples typically involve a retailer or tax firm with a highly seasonal business or a highly targeted business cycle. But is having to manage and drive consistencies all year across the two environments worth it for that short period?
So far, Microsoft is about the only true play here, with its public Azure cloud and private Azure Stack. This strategy gives IT the ability to create an application in a private environment and move to public, or vice versa, relying on the same security, development, libraries and management tools. This is about as close as many will actually come to a hybrid cloud. But, even here, most applications, while they can move from domain to domain, will probably live in one domain or the other 99% of the time.
Time to change how cloud deployments are described
The second group, of which I was a staunch member, simply got the naming wrong. Most businesses will use a mixture of public and private cloud resources, depending on their applications. Those applications with regulatory constraints or data privacy considerations will probably remain in the private domain. This is also where many will see the more differentiated applications for their business — the tools that help define their business and how they deliver products and services to their customers.
Horizontal and nondifferentiated applications will probably move to a public cloud based on economics or a variety of other factors. In the future, most businesses will use multiple cloud resources. This is why hybrid cloud, as a description of this environment, is so wrong. Multi-cloud is a better description. The industry needs to get on board with this.
A great example is in the auto world. If you own one Toyota Prius, you run on electricity or gasoline — a hybrid car strategy. But if you own an all-electric Tesla and a gas-chugging Ford F-150, even though you are using both gas and electric, each is discrete — this is a multi-car strategy.
A complicating factor for a public-private combined entity is the cost and complexity of data transmission. It is expensive and time-consuming to move data back and forth within a hybrid environment. Most businesses are choosing to locate the compute closer to the data and simply send back the results and exceptions, rather than shuttle data back and forth between locations. This dynamic is what is causing a growth of edge computing — moving the compute closer to the network edge, closer to where the data is being collected.
It is time for the industry to change the way it defines hybrid cloud. There will be hybrid clouds, but they will be far fewer and far more limited. The phrase multi-cloud needs to take more prominence in IT’s lexicon, because the need to accurately define the environment will pay dividends in accelerating those plans by removing confusion. The more we rally around nebulous buzzwords and marketing terms, the more we doom our projects to constant challenges and questions. Clarity will help drive better outcomes.