Despite many who predict the death of the data center, the reality is that such a complete extinction is unlikely. A combination of technical, financial, and historical factors point to a future in which public cloud may indeed host the bulk of our applications and data, but there will still be data centers for the foreseeable future.

First and foremost on the list is the need for employees to access applications. Even if every business-related application is “in the cloud,” employees will still need a way to access them – and corporate officers will need a way to govern that access. Whether through secure web gateways, next generation firewalls, or some yet to be introduced technology, IT will need to provide networks and services to the corporate population. This often requires a dedicated data center as business growth – and employee head count – grows. Once a physical data center presence is established, it becomes tempting to “just put it here, for now” and within a few years, voila! Now we’re managing both public cloud and an on-premises data center (again).

Product lifecycles vary by industry. While the lifetime of the latest Internet-enabled gizmo is likely in the two to three-year range, other “products” have a much longer lifecycle often backed by aging, legacy applications. Insurance is a prime example of this. Policies written fifty years ago and supported by applications developed on mainframes cannot easily be extricated and moved to the cloud. Processes and policy-based products – common to the financial industry – are often tied to lengthy lifecycles that make it financially (and sometimes technically) infeasible to “lift and shift” or “refactor” for a public cloud. These applications will continue to reside on-premises, and the existence of such systems often leads to continued development and usage for new offerings. A complete migration to public cloud remains unlikely for such market segments, at least for the foreseeable future.

Security concerns continue to lead the list of reasons organizations are hesitant to go all-in with public cloud. These are not the same concerns business had a decade ago when cloud was a nascent technology. Concerns today are more specific and generally revolve around data, both in terms of consumer privacy and corporate property. Visibility and service capabilities are limited in public clouds, and organizations often require more than a provider is willing (or able) to offer. These restrictions can be the deciding factor in whether an organization invests in public cloud or an on-premises private cloud.

Security concerns are not all technology-related. Public cloud is global, the laws of nations are not. Data privacy regulations vary from country to country, and failure to comply can be a real issue for organizations. Many choose to remain “at home” on-premises rather than struggle through attempting to verify and implement the controls and systems required in a public cloud.

Rarely discussed is the omnipresent reluctance to get “locked in” to a provider. It’s easy to forget, but cloud is a product, and products are built with the hope that someone will adopt it and stay with it forever. This is why organizations prefer standards-based everything — from networking, to security, to application integration. Standards are their means to avoid being subject to the whims of any given technology provider. Most markets have adapted to that reality and are based on standards. They provide “value add” above and beyond to cajole customers into sticking with them. Cloud, today, is a dazzling array of independent offerings, with no standards in sight. None currently enable an easy exit strategy for customers. Once deployed, organizations adopt consoles and application program interfaces (APIs) (and virtualization formats) that are incompatible with other competing cloud providers. They build operational scripts and systems that integrate with that public cloud that leash them even more tightly to the provider. This ultimately means an expensive and painful migration if the organization ever chooses to find a new home. Some organizations count this is a significant deterrent to adopting cloud and turn instead to on-premises options.

The Internet of Things (IoT) is a real technology cycle that’s already begun. As organizations begin to entrench sensors and monitors to digitally transform their business or improve safety and condition of manufacturing equipment, they need networks over which to communicate and systems to collect and display that information. This Industrial Internet of Things (IIoT) relies on speed and security, both of which may not be adequately provided by public cloud providers. Additionally, it makes little sense for organizations to configure sensors on the manufacturing floor to send alerts to the cloud and then back to the people who must react to them. On-premises private cloud provides organizations the control and performance they need to support these systems, and are unlikely to take the risk of employing public cloud when no guarantees (or accountability) exist to assure performance and availability.

Market saturation for most products – technical or not – tends to be reached when about 80 percent of the total addressable market has adopted the product. As an example, virtualization — despite the hype — has never achieved 100 percent penetration. It is at saturation and unlikely to achieve higher rates, especially with containers knocking on every door. Public cloud remains at a low market saturation today and is going to grow. From the data available regarding strategies and plans of organizations, cloud is likely to see 50 to 60 percent of most organizations’ applications, but very few are ambitious enough to claim 100 percent use of public cloud.

There are a lot of factors that go into the decision to take advantage of public cloud or not. Most organizations are adopting cloud on an app-by-app basis, because the decision isn’t about business, it’s about application and data suitability to the environment. And that ultimately means that some apps are going to remain on-premises in a traditional or private cloud model and others will be deployed in a public cloud. Organizations are going to operate in a multicloud, distributed model for the foreseeable future.