The risk is that this can result in an even more complex and confusing jumble of
services. What we need instead is to add these new services and then operate the
resulting infrastructure as a single coherent and adaptive whole – as genuinely
We specify ‘IT’ here because it’s important to remember that this is not just about
cloud. Public cloud is not “better” than the traditional data center, nor is the
reverse true – both have roles to play. Some workloads will not suit the cloud model
at all, while others might fit that model well but require private or hybrid cloud,
not public cloud.
Each platforming decision may involve a range of relevant factors, such as
complexity, business value, security, latency, and so on.
For example, you might consume services of a more generic nature from public clouds,
but have others hosted in some form of private cloud, whether operated from your own
data center or a co-location facility, or run for you by a service provider.
In addition, some applications are becoming ever more dynamic as developers adapt to
a world of continuous change. That means a workload’s requirements may well change
significantly many times during its life, with each change prompting a reassessment
of its location. For example, new usage patterns could make the public cloud a more
cost-effective location, or a new compliance regime could encourage a move back
Hybrid IT might appear complex to implement, but it doesn’t have to be. That is
especially true if you’re establishing an on-prem cloud, either to pull workloads
back from a public cloud or to support your evolution from on-prem IT to a hybrid
cloud model. This is because many of the technologies and concepts that enable cloud
computing also align clearly with integrated systems.
The integrated system is a combination of server, storage and networking that’s
bought, installed and operated as a unified solution. The best-known example of this
approach is probably HCI (hyper-converged infrastructure), but the same parallels
can be seen in converged infrastructure (CI), which is the other main integrated
These two models differ in their physical configuration, but the differences are
largely irrelevant in the hybrid IT world, where the expectation is that the key
resources of compute, storage and networking will be abstracted, and quite possibly
software-defined too. Seen through this lens, integrated systems become a way of
reducing complexity by shifting the focus to where it ideally belongs – the software
With integrated systems, the major decisions and caveats also become software-based.
For instance, which hybrid software stack or stacks will you standardize on, such as
Microsoft, Nutanix, VMware or an open-source option? If you rely on specific
management tools, which hybrid infrastructures and integrated systems can they
connect to and support?
Remember here that you are building hybrid IT for your needs. Of course it needs a
degree of future-proofing – including support for upcoming cloud-native
architectures, for instance, if you are not already using them. But the important
thing is to focus on building the standardized and smoothly-interoperable
environment that your organization needs, and not trying to enable or support
everything under the sun.