The adoption of on-premises clouds has been much slower than public clouds due to a painful realization that clouds are anything but easy to set up and operate. Companies are risk-averse, and complexity brings a lot of risks. Also, the sheer expense of hiring expertise to build and manage on-premises clouds has been daunting for most companies. Now, self-managing clouds, or cloud-managed data centers, are emerging as a solution. Let’s take a quick look at what a cloud-managed data center can offer, how it works, and what the drawbacks may be.

In a cloud-managed data center, a “brain” in the cloud receives telemetry data from on-premises equipment and helps to discover, initiate, configure, or monitor the on-premise asset. Unlike using a managed service provider, the customer relies on the cloud brain to handle the heavy operational lifting. This helps reduce the need for remote sites to have IT workers and drives the idea of centralized management and a single view of all assets. Companies in different segments of the IT market are leveraging cloud-managed architecture for WiFi management and other tasks. Let’s see how cloud-managed architecture can work for private cloud deployments.

There are three groups of challenges to overcome with private cloud:

1. Technical challenges. Technology gets complicated due to extra layers of abstraction and consolidation of resources.
2. Human challenges. It is hard to find experts who understand all the components of a cloud, such as compute, virtualization, storage, and networking.
3. Process challenges. The enterprise must put new processes in place to handle the self-service world of cloud consumption, where developers can upload their own images and create their own networks, switches, routers, software ­firewalls, load-balancers, and storage volumes.

So, if companies aren’t investing in on-premises clouds due to technical, human, and process challenges, what can be done about it? Artificial intelligence (AI) is the key, and it’s moving into many areas of life. We all see the rise of self-driving cars, for example. Cars appear simpler to use and drive every day, and with parking assist and lane departure warnings we have a lot fewer accidents and dings from bad parking jobs.

In the data center, computers can now tell the service people what is wrong. With artificial intelligence, you can replace a lot of human expertise with cloud management. If your IT infrastructure could by managed by the cloud, you could minimize the amount of mundane tasks you are doing and therefore give yourself a more strategic view of your job, or more time for training or planning. A self-driving private cloud takes its inspiration from public cloud vendors who have achieved greater economies of scale via software-driven operations.

In a cloud-managed data center, software is used for self-healing, doing 24x7 monitoring and management, and providing visibility and insights.

Creation – Cloud-managed data centers simplify the creation of private cloud environments and solve the problem of complex and time-consuming hardware deployment, software installs, and con­figuration. Using pre-configured, hyperconverged appliances as the platform for private cloud, the cloud-managed data center eliminates the need to install even one software component or do any confi­guration.

Monitoring – The cloud-managed data center incorporates cloud-based software that does monitoring and self-healing. This self-healing software signifi­cantly reduces the need for experts that a customer would otherwise have to hire and train. Every cloud infrastructure needs a monitoring and operations center, and a cloud-managed data center incorporates a monitoring and operations layer.

Insights – The same data collection layer used for monitoring is also used to generate insights about applications, provide visibility around infrastructure bottlenecks, and help with placement of applications across public and private cloud infrastructure.

The heart of the solution is a cloud brain which is built from a big data cluster to observe and guide cloud decisions. Changed events, statistics, and health checks are relayed up to the cloud brain to do complex event processing, thereby increasing automation levels, improving mean time to recovery, alerting administrative teams to emerging issues, and providing insights about cloud management.

Enterprises benefit from a cloud-managed data center because they can build and operate a private cloud much more quickly and with much fewer human resources. IT resources can be focused on strategic initiatives rather than cloud operations, and it takes far fewer administrators to operate a cloud-managed data center than it does a human-managed data center.

On the other hand, the trouble with AI is the same as its benefit: it takes a lot of decision-making power out of the hands of administrators. AI is only as smart as the algorithms that drive it, and a cloud-managed data center may not reflect the wishes of human managers. Let’s look at a couple of potential scenarios and how they might be addressed.

This is a valid concern for public cloud and software-as-a-service (SaaS) in general. In the case of cloud-managed infrastructure, solution providers should include a graphical user interface (GUI)/command line interface (CLI) for the on-premises software and ensure clients that all of their data is on-premises.

If this is the case, it is likely that your company still uses SaaS products and just does not realize that SaaS products send information all the time. They typically use port 443 HTTPS, and this is widely accepted as a secure connection. So if you use Salesforce.com, for example, you are in fact sending telemetry through port 443.

Like any new idea, it takes time for the rank and file to trust the new model. As SaaS and public cloud services grow, the idea of a distributed architecture becomes more mainstream.

In an era where data center infrastructure is becoming increasingly complex, companies have but two choices: to continue adding IT personnel, or to rely more on automation to manage that complexity. Cloud-managed data centers are in their infancy, but the idea of using cloud brains to simplify on-premises management holds a lot of promise for the future.