Organizations struggling to decide whether they should put their workloads in a private or public cloud should first consider whether they ought to turn to the cloud at all, and then look at a hybrid cloud model.
It might seem like the trend in IT is to move to the public cloud, but this may not be the right answer in all cases, and not all workloads are right for moving to a cloud environment. Many existing enterprise applications are not built for sharing resources in the way that a cloud platform does.
There are important considerations organizations must take under advisement before moving any workloads to a private or public cloud. The first decision must be whether the workload is suited to run in the cloud. If not, then maintaining a physical, clustered or virtual platform is a better bet.
If the application can work in the cloud, then organizations must still decide whether to use a private or public cloud, or both.
Private vs. public vs. hybrid
One benefit of the cloud is sharing resources elastically, so any individual workload can borrow resources from other workloads, but the resource ceiling for private clouds is easier to hit than in a public cloud. For example, a private cloud can share 100 different workloads, whereas a public cloud may share hundreds of thousands or millions. This allows for increased efficiency with resource sharing.
But one major problem with the public cloud is assessing the cost. It used to be relatively easy to calculate public cloud costs, but the major platforms — Amazon Web Services, Microsoft Azure and Google Cloud Platform — have become more complex. They now offer different types of CPU, storage and networks. Organizations must have in-depth knowledge of how each resource works in real life because they each affect how the overall workload operates.
As such, the key is to start with non-mission-critical workloads to see how a public cloud platform operates, as well as how the actual costs work out in real life. By initially choosing relatively basic workloads and working on a short-term, contract basis, organizations can choose and operate different resource pools to see how their application functions with each one.
For example, does solid-state storage provide the uptick in performance the application needs, and at what cost? Does in-memory data analytics give the organization value for its money? Is there a case for applying GPU compute resources to the workload? What happens to archive data, and what are the costs associated with moving it from one region to another, or recovering it to a different platform?
Once an organization has tried a few basic workloads, it can gain a better understanding of how public cloud platforms operate and which configuration options are available. This can help companies make decisions about whether mission-critical workloads should be moved over to such environments.
Even if an existing workload is suitable to move to a public cloud, it’s not a one-way journey. Resource pricing, availability or service levels may change, and organizations need a plan B to roll back any decision that becomes a constraint on how the business operates.
This is where the idea of a hybrid cloud comes in. The real answer to the private or public cloud argument is actually both.
The idea is to create a seamless, logical platform that enables administrators to move workloads to the right platform at the right time. In modern data centers, this is likely a decision of whether a workload should be on the private or public cloud for a period of time. As microservices see wider adoption, it will become more a case of which part of a composite application — one that is built up from a collection of microservices — should be on what platform and when.
As such, any decisions about the cloud that organizations make now must reflect a more dynamic future. Companies must plan for applications to become far more granular and for microservices to be pulled from different parts of an overall cloud platform at different times as business needs, technical availability and financial aspects dictate.