If your IT environment is not running in the public cloud, you are using either an on-premise infrastructure or a private cloud. What is the difference between the two and why is it important?

There is little debate about the definition of the public cloud. The public cloud is the cloud of the hyperscalers. AWS, Google, Microsoft, Oracle and other players offer you customized and flexible services on their infrastructure.

They can purchase scalable storage or computing power and choose between pure infrastructure or more managed solutions. The pinnacle of the public cloud is the serverless offering, where workloads always get exactly the computing power they need thanks to the cloud provider. The draft law will of course follow at a later date.

Share with others

There is one constant: the cloud provider owns and manages the IT infrastructure. Workloads are distributed across the provider’s servers. It provides efficient hardware and sells virtualized computing power to customers. Your virtual machine with four compute cores and 8 GB of RAM can run alongside a competitor’s powerful machine with 128 cores and 256 GB of RAM.

The public cloud offers scalability, flexibility and security, but it also has a downside. Maybe you want more control over your hardware, or the subscription model for your stable workloads is more expensive than you expected. Or maybe workloads need to run locally for performance or security reasons.

Leaving aside the public cloud, you have two options to choose from, the distinction between which is not always clear: private cloud or on-premises. To put the two side by side, it is important to agree on the definition.

On site

Local, often abbreviated as On-Premliterally means “on-site” and refers to a company’s own location. Sometimes organizations still mistakenly use the term “on site’ without ‘S’but this is wrong and makes no sense. On-premise infrastructure can be considered as a counterpart to public cloud infrastructure.

On-prem hardware is usually located in the company’s own data center. This could be a large server room, a dedicated data center, or a broom closet. You can also rent your own space, for example in a colocation data center. The colocation specialist takes care of powering and cooling the hardware, but we still consider this as on-premises.

In any case, the organization has full control over the hardware. Servers are ordered and assembled according to the company’s wishes. It is up to the company’s IT team to ensure that sufficient storage and processing power is available for the company’s needs.

An on-premises infrastructure is often synonymous with a more classic IT environment in which different servers support the specific workloads of all. Even more modern hardware implementations such as hyperconverged infrastructure (HCI) are possible in the on-premises context, provided the hardware is managed by the company itself.

Private Cloud

Just like on-premises, private cloud is a counterpart to public cloud, but the approach is different. With on-premises solutions, the focus is on the infrastructure, its provision and management. The term private cloud refers to the flexibility of the offering, which is comparable to the public cloud.

Just like in the public cloud, in the private cloud you can allocate virtual machines according to your needs and assign more or less power to workloads. The main difference is that the underlying infrastructure is not shared with other customers, but is – as the name suggests – private. No other customers in the data center can access the storage, compute power or network infrastructure of a company’s private cloud.

There are various ways to set up a private cloud. You can do this yourself, with your own hardware in your own data center, or with a colocation specialist. In this case, the private cloud runs on-site and both come together.

You can also partner with a hardware manufacturer that offers a cloud-like subscription model for servers, where you rent them as you use them but don’t share them with anyone else. Lenovo TruScale, Dell Apex and HPE Greenlake are the most well-known examples.

Finally, you can work with a private cloud provider in the same way as you would with a public cloud provider. The main difference remains that the entire physical IT infrastructure is exclusively available to you. Large cloud providers such as Azure, AWS and Google also offer private cloud offerings in addition to their public offerings. The functionality is similar, but access to the infrastructure remains private to the customer.

Who owns it and what do you share?

As you read, on-premises and private cloud are not side-by-side, but overlap. On-premises refers to managing and owning the infrastructure. You can run a private cloud on on-premises infrastructure, but you can also opt for an old-fashioned three-tier approach where storage, networking, and compute are separately purchased building blocks.

With an on-premise infrastructure, you own your servers: you choose the hardware and decide how modern or classic the infrastructure behind your IT environment is. The payment model is secondary. If you want, you can also rent servers, provided they are located at a location of your choice and you have full access.

With a local infrastructure, you own your servers.

Private cloud is a model that relies on the IT infrastructure. Where the private cloud runs and who owns the hardware does not matter much. What is important is that the private cloud runs on servers that are exclusively assigned to the customer, in a network that has been set up exclusively for the customer. If you run a virtual machine with eight processing cores on a server with 64 processing cores, the remaining 56 cores are available to you. You can deploy them at any time and they will never be assigned to another company’s VM.

Flexibility at a price

A private cloud works just as flexibly as a public cloud. For example, you can allocate available computing power and storage for workloads through a cloud portal without having to dedicate an entire server to a task. You can also allocate more or less power or storage to a workload as needed, just like with the public cloud.

The big difference: Since the servers behind the private cloud are reserved exclusively for you, this computing power has to be ready. Do you suddenly need 24 additional cores and 64 GB of RAM? Then it has to be ready. In other words: You have to have bought them (on-premises) or requested them from your private cloud provider. This takes longer than in the public cloud, where the public cloud provider always provides available computing power.

Since you have to provide this growth space yourself, you have to over-provision a bit. You not only pay for what you use, but also partly for the hardware you hold back for your growing consumption.

If you choose your own hardware, either through a private cloud agreement or because you keep everything on-premises, you need to take this into account. Scaling is possible, but not instantly like in the public cloud. In the pure on-premise context, it all depends on delivery time. If you work with a private cloud partner without pure on-premise infrastructure, scaling can happen faster.

When and why?

When do you choose the public cloud, when do you choose a private cloud and when do you think about a local infrastructure?

Public Cloud

• A model for IT consumption
• Complete hardware security
• You configure what you need
• You share the environment with other customers in a secure way
• Always via an external data center

Private Cloud

• A model for IT consumption
• Works similarly to the public cloud
• The underlying hardware is private
• Workloads also scale flexibly, provided the necessary hardware is available
• You can provide the hardware yourself (on-premise) or work with a partner
• Can be performed locally (low latency) or in an external data center

On site

• A model for IT hardware management
• You place your own hardware at a location of your choice
• You are responsible for the availability of IT capacities
• You are completely free to choose how you use the hardware on site (3-tier, HCI, private cloud…)
• Often in your own data center (low latency), but colocation is also possible

The public cloud therefore remains a smart choice for unpredictable workloads that need to scale quickly. Private cloud is a model that tries to mimic the convenience of the public cloud, but in a private context. Since the hardware is not shared, this model theoretically offers more privacy and security. This can be important for parties in the financial sector or healthcare that need to adhere to strict rules.

If you choose a private cloud from an external provider, the difference with the public cloud is actually not that big anymore. Virtualized workloads are separated in the public cloud in such a way that the security risk is theoretical for most types of customers.

On-premise means self-management of the hardware. The advantages are greater there. If you know how much IT capacity you will need in the next few years, you can purchase it in reasonable quantities. This is often cheaper than running a stable workload with a public cloud provider.

On-premises infrastructure also offers the benefit of lower latency, especially when the data center is very close to the organization. For example, an animation studio might want instant access to huge files without having to send them back and forth over the public internet.

A little bit of everything: Hybrid Cloud

So what is the best choice? This varies from company to company and even from workload to workload. First the cloud is a marketing slogan we are fortunately hearing less and less. Some workloads benefit from scalability, others from flexibility but have to comply with complex regulations, and still others are stable and simply available as close to the end user as possible.

The perfect solution is often a combination of one or more public and/or private cloud providers and a local infrastructure: the hybrid cloud.