How does a private cloud work?
A private cloud works by allocating dedicated servers, storage and networking to a single organisation, then dividing and managing them as a flexible pool of resources.
It is built in stages. Servers, storage, and networking are set aside in a physically isolated environment, so they are used for one customer’s workloads alone. Virtualisation then divides the physical servers into virtual machines (VMs), allowing several workloads to run separately on the same hardware. A management platform allocates resources to those VMs, and to containers, and lets them be adjusted as demand changes. In a managed private cloud, the provider’s engineers then monitor, maintain and support the environment, leaving internal teams free to concentrate on the applications running on it.
Because the hardware serves one organisation, workloads do not compete with those of other customers for processing power, memory or storage performance.
What technology does a private cloud run?
A private cloud is made up of a few core technologies that work together. Kernel-based Virtual Machine (KVM), an open-source hypervisor built into the Linux kernel, divides each physical server into VMs, each with its own operating system and applications, kept separate from the others. In a private cloud all of that hardware belongs to one organisation, so KVM divides resources between the customer’s own workloads, such as production and development environments.
Red Hat OpenShift sits on top as the platform that manages those workloads. It is an enterprise Kubernetes platform, adding built-in security, lifecycle management and tooling to the system that automates how containers are run, and its virtualisation feature, powered by KubeVirt, runs VMs alongside containers. Existing applications can stay on VMs while newer ones use containers, with one place to monitor, allocate and manage resources.
OpenShift is supported by IBM, which puts the backing of a major enterprise technology vendor behind the platform. It is built for agility, security and consistency, so resources can be adjusted as requirements change and the same tooling applies across the whole environment.
How is a private cloud designed?
A private cloud is designed around your workloads, with no fixed template to fit them into.
Our engineers start with a consultation to understand your infrastructure and operations, and the specific requirements of your applications, then choose the hardware, platform, network, resilience and security to match. Typical questions cover what will run on the platform and what resources it needs, how much demand will grow and when the peaks fall, how much downtime or data loss each workload can tolerate, how sensitive the data is and which certifications apply, where your users are located, and how much you want to manage yourselves. For example, Spearhead AG, an insurance technology provider, needed isolated hardware, an ISO-certified hosting provider and a certified Platform as a Service (PaaS) environment, and its platform was designed around those requirements.
Not every workload needs the same level of resilience, so these questions are often asked application by application. A business-critical database may justify replication to a second location, while a development environment may not. Location is part of the design as well, since platforms can be deployed in more than 35 locations worldwide, chosen to suit latency and data requirements.
What does a private cloud environment look like?
Every private cloud environment is designed around its workloads, so no two look the same. A typical web application, as shown in our Build Your Perfect Cloud configurator, uses an OpenShift VM stack with a firewall, load balancer, web tier, database, backups, a Cloudflare content delivery network (CDN) and an optional disaster recovery (DR) mirror of production, all running on servers and storage used only by one customer.
Following a request through that environment shows how the parts fit together. The CDN delivers content from locations closer to users, then a firewall controls the traffic allowed through and a load balancer shares requests between several web servers, so no single server is overloaded. The web tier VMs run the application and call on the database, a separate set of VMs that holds the data, while backups take regular copies for recovery and an optional DR mirror keeps a copy of production at a second location. Other workloads, such as ERP systems or large databases, need different arrangements.
How does a private cloud scale, stay available and stay secure?
The way a private cloud is built determines how well it handles growth, failure and risk.
Scaling on demand
Resources adjust on demand, with you in control of the allocation of resources. Due to the platform being virtualised, CPU, memory and storage can be added or taken from VMs, and VMs can be created or retired, without buying new hardware each time. On OpenShift, containers scale the same way as VMs. Scaling happens within the capacity of the platform, which is planned in advance, depending on the needs of your business.
Planning is what makes private cloud work. Remarkable Commerce, an eCommerce platform provider, previously added new dedicated servers every year to cope with Black Friday. After moving to a private cloud, it scaled resources up and down around peak times, with stress testing beforehand and alert thresholds set so that extra resources could be allocated in the event of a surge in traffic. In 2022, with 43% more new users than the year before, it eliminated virtual queues and saw successful transactions rise 89% compared with the three weeks before the event. Think Learning, an eLearning provider that grew rapidly during the pandemic, moved to a bespoke private cloud with the power its learning management system requires, migrating its existing client base without disruption.
Staying available
A private cloud can be built with redundancy at each layer, so the failure of a single component does not take the service down. Replication to a secondary data centre and regular backups protect against layer failures, and an optional DR mirror can take over if the primary site becomes unavailable. Backups and replication work together, since replication follows changes as they happen whilst backups provide earlier points to restore from.
Staying secure
Because the hardware is dedicated and no resources are shared with other customers, the platform is isolated from other workloads at the physical level, with security layered on top. Compliance requirements, such as those in regulated industries, can shape the design, from certified hosting to where data is located. Where data is hosted and which laws apply to it are separate questions, so both are worth considering. Our UK sovereign private cloud is designed for regulated environments, keeping data in the UK under UK jurisdiction.
How is a private cloud migrated and managed?
Moving to a private cloud starts with the consultation and design described above, followed by a staged migration that prioritises live sites and applications, keeping disruption to a minimum. Once the platform is live, the provider continues to monitor, maintain and support it as requirements change.
With our managed private cloud, customers have direct access to a dedicated technical support engineer 24/7/365, with no tiered support, and our team acts as an extension of their business. On our HPE Synergy servers, we guarantee hardware replacement within 20 minutes of a hardware failure under our service level agreement (SLA).
What should you consider before choosing a private cloud?
Choosing a private cloud involves more than selecting hardware. The wider design and operating model should also be considered.
Questions worth asking include:
- How much compute, memory and storage will your workloads require over the next three to five years, and how quickly might that change?
- Do your applications handle sensitive or regulated data, and where must it be hosted?
- How much downtime or data loss can each workload tolerate?
- What backup and disaster recovery requirements exist?
- Who will manage updates, security patching and monitoring?
- Does your provider offer proactive monitoring and rapid hardware replacement if components fail?
Considering these requirements early helps ensure the platform can support both current workloads and future growth.
The next steps
A private cloud works by providing dedicated hardware, virtualisation and a management platform, designed around your workloads and run by a provider that monitors and supports it. For workloads that need consistent performance, strong isolation and room to scale, it provides a flexible solution that can grow with your business.
Understanding how a private cloud works is the first step. Choosing the right design, location and management model requires a broader assessment of your applications, compliance requirements and plans for growth.
If you are evaluating a private cloud, speaking with an experienced managed hosting provider can help determine whether it is the right fit for your workloads and how it should be designed to support long-term objectives. You can explore the options in our configurator, or speak to one of our Cloud Consultants for an initial conversation.
