What to keep off the cloud: the business systems that still belong on site

Cloud adoption is not an all-or-nothing decision, and the businesses that get the best results from it are usually the ones that kept a few things deliberately on site. This article sets out which business systems still earn their place on local hardware, why, and how to run them alongside cloud services without ending up with the worst of both.

The short version: cloud by default, local by exception

For most businesses of up to 250 people in Hertfordshire and London, email, documents, collaboration, accounting and CRM belong in the cloud. The reasons are well rehearsed: no server to maintain, access from anywhere, predictable per-user cost, and resilience that a small business cannot build itself.

The exceptions are systems where one of five conditions applies:

1. It must keep working when the internet connection does not.
2. It needs response times measured in milliseconds, consistently.
3. It is physically wired to machinery, doors, cameras or sensors.
4. It moves very large files all day.
5. A regulator, contract or insurer has imposed specific conditions on where the data lives.

If none of those apply, the case for keeping a system on site is usually habit rather than need. If one or more applies, the answer is normally hybrid: local for the function itself, cloud for backup, reporting and remote access.

Systems that must survive an internet outage

Cloud providers publish high availability figures, but availability at their end does not help when the fibre to your industrial estate is cut by a contractor. Some functions are too important to depend on the connection.

• Door access and alarm systems. Staff need to get into the building at 7am regardless of what the broadband is doing.
• Warehouse, yard and depot operations. Barcode scanning, hire-out and return, weighbridges and stock movements should record locally and sync later.
• Point of sale and till systems. Offline-capable is the minimum requirement.
• Local authentication. If every login depends on a cloud identity service and the line goes down, nobody can sign in to anything, including the systems that do not need the internet.

A second connection with automatic failover, covered under Dig IT's multi-site connectivity service, reduces the risk for everything else. It does not remove the case for keeping these particular functions local.

Latency-sensitive applications

Every cloud round-trip adds delay. For a Word document nobody notices. For a database-driven application where each keystroke or barcode scan queries a server, tens of milliseconds per operation add up to a system that feels broken, and staff will find workarounds that create bigger problems.

Typical examples in the businesses Dig IT supports:

• Hire management, job costing and stock control systems built on a local SQL database.
• CAD and design applications working on large models.
• Practice management systems with heavy database traffic.
• Manufacturing execution and production control software.

Some of these have cloud-native versions now, and moving to one can be the right long-term answer. Lifting an application designed for a local network onto a cloud server without redesigning it, however, usually produces slow software and an expensive monthly bill. Test with real users and real data before committing.

Anything wired to physical infrastructure

Machinery, building management, CCTV recording and industrial sensors talk over local networks with protocols that were never designed for the public internet. The control loop needs to be on site. What belongs in the cloud is the layer above it: dashboards, alerts, remote viewing, and long-term storage of recordings and logs.

CCTV is a good illustration. A UniFi Protect system, as installed by Dig IT at the Hertfordshire Golf and Country Club, records locally to an on-site appliance so footage is captured even if the connection drops, while remote viewing from a phone works through the cloud. That split is the right model for most physical systems.

Very large working files

Video production, architecture, engineering, print and photography businesses routinely work with files measured in gigabytes. Editing those directly from cloud storage over a typical business broadband connection is painful, and syncing every version through OneDrive consumes bandwidth the rest of the office needs.

The usual answer is a fast local NAS, such as a Synology unit, on a properly cabled network for day-to-day work, with the cloud used for finished deliverables, sharing with clients and off-site backup. Dig IT's structured cabling work exists partly because a 10 gigabit link to a local NAS solves a problem that no amount of cloud subscription can.

Sites with poor or restricted connectivity

Construction sites, rural depots, temporary premises and secure facilities may have slow, unreliable or deliberately restricted internet. Systems that must run at those locations, such as site safety logging, local data capture or plant telematics gateways, need to work standalone and synchronise when they can.

Cloud applications with an offline mode often cover the basics but with reduced functionality and delayed synchronisation. For anything safety-related or revenue-critical at a poorly connected site, design for local operation first.

Regulated data: a documented decision, not a ban

UK GDPR does not prevent cloud use, and the ICO's guidance for organisations is clear that accountability stays with you whichever platform you use. Major providers offer UK data residency and publish the contractual and security commitments needed for a defensible decision.

Where businesses do need care:

• Government and defence contracts that specify hosting, personnel or location conditions.
• Professional regulators with rules on client records, retention and access.
• Insurance conditions that reference specific controls or data locations.
• Long-term archives where the retention period is measured in decades and cloud subscription cost accumulates for the whole period.

The NCSC's cloud security guidance gives a framework for assessing a provider. Use it, record the decision, and revisit it when contracts change. Keeping regulated data on site is a legitimate choice, but it makes you responsible for physical security, patching, backup and access logging that a provider would otherwise handle.

How hybrid works in practice

Mr Plant Hire, the Enfield-headquartered plant and tool hire business Dig IT has supported for over 15 years, shows what a well-designed hybrid looks like. Multiple depots are linked by site-to-site VPNs. Windows servers on site run the core business systems where speed and resilience matter, protected by both local and cloud backup. Microsoft 365 and SharePoint handle email, documents and inter-company sharing across the group, so staff at any depot see the same files without the servers being involved.

The principle is simple: put each workload where it works best, connect the two halves securely, and back everything up in more than one place. Dig IT's cloud services and backup and disaster recovery teams design around that principle rather than around a preference for either cloud or on-premise.

A quick test for any system

Before moving or keeping a system, answer these four questions honestly:

If every answer is no, the system is a cloud candidate and keeping it on site is probably costing more than it saves.

What to do next

If you are unsure which of your systems genuinely need to stay on site and which are there through habit, a short review of your applications, connectivity and backup arrangements usually settles it. Dig IT offers this as part of an IT health check for businesses across Hertfordshire, west Essex and London.