Zootnet Ltd - An IT Architectural Practice
On-Premises
Just because new and better ways of deploying technology become available - that doesn't mean they can be exploited straight away. It takes time to untangle a business function from a legacy system. Fortunately, here at Zootnet Ltd we learned our trade on those legacy systems. That's why we're ideally placed to best understand the processes of migration and whether there is any benefit in doing so.
On-premises is a deliberate choice, not a leftover
Public cloud is the right home for many workloads, and our cloud guidance explains why. But cloud adoption is a decision made workload by workload, and for a significant number of systems the best answer is still to run them on infrastructure you control. In most cases the strongest architecture is a hybrid one: local where the risk, safety or physical constraints demand it, and cloud wherever it genuinely adds value.
Modern on-premises is also not what it was. Hyper-converged infrastructure, containers and cloud-managed local platforms let you keep the control of your own facility while gaining automation, patching, monitoring and policy from the same tools used in the cloud.
Where on-premises remains the right model
Regulated and safety-critical sectors show the pattern clearly. The regulators in these sectors rarely ban cloud outright. They require that risk is assessed, assured and owned. Where the assessment shows that control, locality or resilience matters most, on-premises architecture is the sound outcome.
Emergency services and policing
Control rooms, dispatch and mobile data underpin life-critical decisions. Availability, assured physical security and clear accountability for law enforcement data often outweigh the convenience of a shared platform.
- Command-and-control and dispatch systems need predictable latency and must keep working when a wide-area link or a cloud region is unavailable.
- Police information is subject to national policing security policy, and the Police-Assured Secure Facilities (PASF) scheme requires physical inspection of the facilities that hold sensitive data.
- Critical communications such as Airwave, and its planned replacement the Emergency Services Network, run alongside on-site control room infrastructure for years, so integration with local systems is unavoidable.
- Forces increasingly use public cloud for suitable workloads, so the realistic target is a hybrid estate where each system sits where the risk, resilience and assurance evidence support it.
NHS, hospitals and clinical systems
Patient safety, long equipment lifecycles and strict information governance mean that many clinical systems remain on site, or on local infrastructure that is managed from the cloud.
- Imaging modalities, theatre, pathology and other departmental systems depend on specialist, validated hardware and local networks, and are often certified as medical devices.
- Clinical safety standards DCB0129 and DCB0160 require manufacturers and deploying organisations to manage clinical risk through the lifecycle, including upgrades. This favours stable, well-understood platforms.
- NHS guidance permits public cloud for health and care data, but places the risk decision with each data controller. Some data and services will still pass a risk assessment better when held locally.
- Wards and clinics need to keep working through a network outage, so local resilience is a patient safety requirement, not an optional extra.
Essential services and critical national infrastructure
Energy, water, transport and other operators of essential services are assessed against outcome-based frameworks that place a premium on resilience, isolation and knowing exactly where systems run.
- Operational technology and industrial control systems are frequently air-gapped or tightly segmented, and cannot tolerate unplanned reboots or forced upgrades.
- Outcome-based assessment, such as the NCSC Cyber Assessment Framework, expects clear ownership of essential functions and tested recovery, whatever the hosting model.
- Sovereignty, classification and contractual requirements can restrict where data and administrative access may sit.
- Public bodies and defence suppliers may hold material that policy keeps within controlled facilities.
Other good reasons to stay on site
Data residency and sovereignty
Some contracts, regulators or classification schemes require data, keys and administrators to remain in a defined jurisdiction or facility. Local hosting makes that simple to evidence.
Latency and locality
Factory floors, theatres, control rooms and retail sites need a response in milliseconds and must keep trading when the wide-area link fails.
Legacy and specialist workloads
Mainframes, bespoke applications, licensed software tied to hardware, and equipment with long certified lifecycles are often safer to leave in place until a planned replacement.
Predictable, steady cost
Stable workloads that run at high utilisation for years can be cheaper on owned or leased infrastructure than on pay-per-use pricing, particularly with heavy data egress.
Control of change
Validated systems need change windows and testing that suit the business. On-premises lets you control when, and whether, a platform changes underneath you.
Disconnected and edge operation
Ships, remote sites and secure facilities may have little or no connectivity, so the platform must operate fully on its own.
Modern on-premises hosting models
| Model | What it is | Typically used for |
|---|---|---|
| Traditional servers and virtualisation | Physical servers and hypervisors managed by your own team. | Established line-of-business systems and specialist applications. |
| Hyper-converged infrastructure | Compute, storage and networking combined into scalable, software-defined clusters. | Consolidating datacentres and clinical or control room platforms. |
| Containers and local Kubernetes | Portable workloads running on clusters in your own facility. | Modernised applications that must stay local but benefit from DevOps. |
| Cloud-managed local platforms | Provider-supplied stacks such as Azure Local, AWS Outposts or Google Distributed Cloud, hosted in your site. | Keeping data local while using familiar cloud tooling and management. |
| Edge and disconnected | Small, self-sufficient systems at remote or isolated sites. | Ships, remote sites, field operations and secure facilities. |
How we help you decide
- Classify the workload. What data does it hold, which regulations and standards apply, and what is the impact of failure?
- Test the assumptions. Compare real costs, resilience and assurance evidence for on-premises, IaaS, PaaS and SaaS options.
- Plan the exit routes. Even if a system stays on site today, design interfaces so that it can be replaced without disturbing everything around it.
- Connect it properly. Local systems still need secure, well-governed links to cloud and SaaS services. See our integration guidance.
Public guidance, standards and documentation
These are publicly available sources that we use when assessing regulated workloads. They are written by regulators, public bodies and vendors, so read them alongside each other. Standards change, so always check the date and version.
Healthcare
National guidance and standards for health and social care IT in England.
Policing and emergency services
Public material on policing security and critical communications.
Security and assurance frameworks
Outcome-based frameworks that apply whichever hosting model you choose.
Hybrid and local cloud platforms
Vendor documentation for running cloud-style services in your own facilities.
External links open in a new tab. Zootnet Ltd is not affiliated with these organisations and is not responsible for their content. This page is general information, not assurance or legal advice for your organisation.
Not sure what should stay and what should move?
Zootnet Ltd can assess your estate against the risks, standards and costs that apply to you, and design a hybrid architecture that fits.