Up to now, establishment  of all magnitude have been applying a siloed operations model to build out data center infrastructures that serve a definite function or run a definite application. This “one tool, one job” has made many of the big, advanced technologies unavailable to small and midsize companies simply because they couldn’t justify the cost.

2019 is believed to usher in  maximization of multi- purpose data centers  It’ll also be a year of converging concepts that data center infrastructure has, until now, flirted with or struggled to wrap into a single offering. Federated architecture models, single-click compostability, advanced automation of mundane IT tasks and new levels of intelligence that’s predictive rather than simply reactive.

Presently, we categorize the core data center, edge and cloud as distinct environments that all need management tools and on-site utilization plans. But a combination of things will happen that will bring these different technologies into one infrastructure. First, businesses will continue to become more comfortable with on-premise private clouds. They’ll better understand the importance of edge computing. And they’ll also better see how the public cloud can help fill the gaps, whether it’s for burst capacity, for specific workloads or to scale up and down for other needs. At the same time, we’ll see faster pipes, better compression and enhanced security, all making this public cloud a more realistic extension of your on-premises infrastructure. It’ll allow the edge and the core to start operating as federated systems and burst out to the cloud when capacities require it.

But it won’t only be an innovation in architecture, but in infrastructure management as well, where all rich data services and data center resources—no matter their location or resource type—are pooled together and managed in a single location with a single GUI. These resources will also be managed with advanced intelligence and by policies that determine which workloads need what resources and where it’s best for them to reside to optimize I/O operations and costs, as well as to guarantee SLAs. All of these features will take a load off administrators.

Will we get there in 2019? Not all the way, but we’ll see major steps toward sharing some of the common tools across the various locations where previously you’d consider them compute islands. You’ll start to operate as a federated infrastructure—though maybe not yet as a shared pool. But the same tools can manage all those locations and how data and applications move between them.

This move to a federated infrastructure is a core part of integrating IoT into your IT systems because more composability allows IT to orchestrate all of these resources through a single, easily programmable API. Doing so is ideal for large-scale mixed-workload environments, and IoT is by nature a very mixed-workload environment. IoT adoption is on the rise: 80 billion connected devices by 2020, 165 zettabytes of data generated per year. In 2019 we’ll see vendors making great strides to ensure advances in composable infrastructure serve this unique nature of IoT.

People are becoming more sophisticated. They’re now realizing they can reconfigure infrastructure on the fly to support applications based on the needs at that moment. They don’t have to provision every application with its own infrastructure. The year 2019 is showing positive signs of being when composable infrastructure takes off.

Consider the example of an expanding airport I worked with recently, located in a high-growth region. It sounds odd, but the airport is landlocked. So, when leaders there wanted to bring in more flights, they lacked room to build more gates. The problem: all the current gates were already leased, but they weren’t all in use continually. If leaders could allow other airlines to use the gates when the primary lessees weren’t, they could bring in more flights without expanding the airport’s footprint. Dynamically leasing the gates solved one problem, but it created another. Right now, each gate is provisioned with IT infrastructure to support the main lessee’s applications.

With composable infrastructure, however, this airport is dynamically reprovisioning the resources required to support those applications. Now they can support the secondary carriers’ needs and dynamically build a new infrastructure to support the gate’s current flight. This kind of composability will accelerate throughout 2019 with continued innovation in sophisticated features and capabilities that provide rules-based reprovisioning of resources in real time according to application need.

Machine learning and artificial intelligence have been on everyone’s list of near-future technologies for a while, and with good reason. When you look at the leading edge of what these technologies can do, it’s impressive. I don’t see that changing any time soon, and I think this coming year will see real progress toward AI becoming self-sustaining as opposed to needing a human hand to guide it.

You’ll see machine learning and AI translate into technologies such as self-healing systems—predictive systems for ensuring systems don’t fail. You’ll see the technology gain the ability to intelligently assess capacity as well as performance. That technology will then automatically optimize infrastructures to increase performance and efficiency.

For example, by using machine learning and AI, the system will understand where it’ll need more capacity and then go and reserve that capacity from wherever makes the most sense. Maybe it’s from the cloud because the need for more storage is temporary. Or maybe the system creates a manifest for the administrator to use when buying more storage for the physical data center because it’ll be necessary in the next three months. AI and machine learning will become more effective at actually managing your infrastructure for you and will provide true intelligence on how to plan. It’ll be predictive, not just reactive.

How this feat is actually accomplished may be unimportant. It could involve stochastic modeling. It could involve basic trending. It could even involve crowdsourcing: “Okay, here’s our experience with other customers, so you should expect the same sort of thing because your workloads look similar.”

Because more data is being shared with vendors on how customers operate their infrastructures, we can start to compare what other customers are seeing and experiencing and use that insight to do predictive modeling. So, when heat spikes in a particular set of servers you can anticipate they are either failing or exceeding the cooling capacity. That insight might prompt moving stuff off site. It might require a change to the infrastructure. It might require simply balancing the workloads to minimize heat generation on location. Those needs we can sense and resolve immediately with this next generation of machine learning and AI—as opposed to just turning on a red light on the dashboard and making some human correct it.

Technology moves fast, and those with deep pockets will always have greater access to the best technology. But as we move into 2019, the increased accessibility of these cutting-edge innovations will lead to greater efficiencies and better outcomes for IT and business.