Cloud WMS vs on-premise — what to choose for an SMB

Andrzej Lenkowski,

Five years ago the question "cloud or own server" in a European SMB usually had one answer: "own one, our data has to stay with us". Today the split is different - around 70% of our new implementations go to the cloud, the rest is on-premise or hybrid. What changed mostly is that connections got more stable, GDPR set clear rules, and the cost of own servers in a five-year refresh cycle turned out to be not so low. But there are cases where on-premise still wins. Below a few specifics.

On-premise WMS in practice

On-premise means: a server physically at your site or in a rented data centre, run by you (or your IT company). The client has SQL Server on a machine in the server room, the WMS application installed on it, and terminals connect over Wi-Fi within the warehouse. On top: backups (usually an external disk plus cloud as a safety copy), monitoring (someone has to watch that the server came back after a power cut), updates (you roll them out and you test them).

Pros: full control over the data, no dependency on an external link, low local network latency. Cons: cost of hardware and admin, responsibility for downtime, the need to keep updating - leave the version two years untouched and you stop on outdated technology.

A classic case from our practice: a pharma distributor near Warsaw, industry regulation forces full data control, the data is critical and expensive in case of a leak. On-premise is the only variant that passes the audit. But that is a specific scenario, not the rule.

Cloud WMS (SaaS)

SaaS means: server at the vendor, access through a browser or a dedicated app, data in their data centre (typically AWS Frankfurt, Azure West Europe, OVH, or the vendor's own DC). You log in, you work, everything runs "from the cloud".

From the user's point of view: nothing in the basement, installing on client machines is a matter of minutes, a new system version arrives by itself and usually doesn't need any client action. Updates, security patches, backups, monitoring - that is the vendor's work.

Pros: zero infrastructure management, automatic updates, access from anywhere, scaling up and down without discussion. Cons: dependency on the link, a monthly fee with no end, less customization flexibility in some SaaS models.

Costs over 5 years - a concrete comparison

A medium warehouse, 8 users, ERP integration, basic customizations. We are counting in EUR.

Cloud (SaaS), 8 sessions × 90 EUR/month. Annual subscription: ca. 8,600 EUR. One-off implementation: 5,500 EUR. Year-one total: 14,100 EUR. Five years: 5,500 + 5 × 8,600 = 48,500 EUR.

On-premise, perpetual licence. Licence for 8 workstations: 11,000 EUR. Server plus Windows Server / SQL Server licences: 4,000 EUR. Implementation: 5,500 EUR. Annual maintenance 20%: 2,200 EUR. Five years: 11,000 + 4,000 + 5,500 + 5 × 2,200 = 31,500 EUR.

Looking at these numbers alone - on-premise comes out 17,000 EUR cheaper. Except you have to add things that never appear in the on-premise quote: half an FTE of IT admin over 5 years (at 1,800 EUR/month that is 54,000 EUR, but it usually shares with other tasks, so the real net add is 7,000-13,000 EUR), server refresh in year five (2,500-3,500 EUR), power and cooling (500-700 EUR per year). Once those are in, on-premise and cloud both land in the 45,000-55,000 EUR area - the gap becomes symbolic.

Which means: over a 5-year horizon the decision is not about price. It is about who runs what.

Speed of deployment

Cloud: the environment is ready an hour after signing the contract. The consultant logs in, configures the structure, the client starts training in week two. Realistically, contract to go-live for a small business - 3-5 weeks, for a medium one - 8-12.

On-premise: you have to order a server (2-4 weeks delivery), install it, configure it, and only then put the application on top. Plus possibly a warehouse network audit, access point installation, opening firewall ports. For a medium business 4-6 months is nothing unusual.

For a client who wants a WMS "in time for the holiday season" or "in the new warehouse opening in three months", cloud is often the only realistic option.

Reliability and uptime

A standard SLA at a serious SaaS vendor is 99.9% (around 8 hours downtime per year) or 99.95% (around 4 hours). A qualified team is doing it 24/7 for you.

On-premise: your SLA is whatever you have internally. The server hangs Friday evening, the admin is off work - no one will fix it until Monday. Disk replacement, power outage, room overheating - these scenarios actually happen at our clients several times a year. An SLA of "call us, we will be there in 24 hours" sounds calm until those 24 hours are spent with the warehouse standing still.

In practice: on-premise clients have more short downtimes than cloud clients. Bigger cloud outages (e.g. AWS Frankfurt down for three hours) happen twice a year on a global scale - and they make the news. Local outages at the client - daily life, which no one counts.

Data and security

The myth "my data is at my site, so it is safe" still lingers, but the truth is often the opposite. In the cloud, data sits in a certified data centre (ISO 27001, SOC 2, typically TIER III or IV) with physical monitoring, fire suppression, redundant power, geographically distributed backups. At the client it sits on a server in a room with a wall AC unit, a key in the door and a cleaner who yesterday unplugged the UPS to plug in the vacuum cleaner. That is not a joke - we have seen it.

GDPR: both models comply if the vendor has a Data Processing Agreement (DPA) and processes within the EEA (or in a country with an adequacy decision). Check the vendor's data centre - most serious SaaS providers host inside the EU and present a DPA on request. In the cloud you usually also get ready audit reports.

Backups: in the cloud usually automated and outside your responsibility. On-premise: you do them yourself. The most common on-premise failure we have seen - "yes, we run backups", but no one has tested a restore in two years. A backup that doesn't fire on a real failure is not a backup.

Offline work in the warehouse

The classic argument for on-premise: "we have no Wi-Fi in the freezer". A fair argument, but it doesn't decide the case.

First: cloud WMSes with a well-designed mobile app have an offline mode. Operators scan, documents buffer on the terminal, then sync with the cloud once Wi-Fi is back. In Weaver WMS this is a standard feature - operators don't notice the difference between full coverage and a shadow zone, provided the sync closes before end of shift.

Second: the question isn't "cloud or on-prem", it is "do I have Wi-Fi in the freezer". Because on-premise with the server in the office and the warehouse 50 metres away with no network will behave exactly the same as cloud. Internet for cloud and Wi-Fi for on-prem are the same physical problem.

Third: in warehouses with extremely unstable links (mobile, super-localised) hybrid sometimes makes sense - a local intermediate server that syncs to the cloud when the link is up and buffers when it is down.

Integrations and customizations

Here on-premise sometimes wins: full control over the database, free hand to write your own pieces of code, integration with a local ERP through the SQL database without exposing an API to the outside.

SaaS has to deliver the same needs through REST API, webhooks, sometimes GraphQL. That is not worse - it is simply a different way. Each new integration goes through a clear, documented interface. In practice a new cloud integration takes comparable time to on-premise, sometimes faster.

Look-and-feel customizations, additional fields, reports: in SaaS usually through a configurator. Fundamental customizations (entirely new modules) - some SaaS vendors allow them, some don't. Check what exactly "customization" means in the cloud offer you are considering.

Hybrid - when it makes sense

Some implementations are mixed: part runs locally, part in the cloud. The two most common variants:

  • Local edge server + sync to cloud. A small machine in the warehouse buffers terminal operations, at night everything reconciles with the cloud. Good for warehouses with shaky links or with strict low-latency requirements within the local network.
  • On-premise main database + cloud reporting layer. An old on-prem WMS keeps running, but you want dashboards available to the management from anywhere. Data flows to a cloud warehouse once a day, into Power BI / Looker.

Hybrid is more expensive than the pure models (you pay for hardware and for the cloud), but sometimes it is the cheapest one in business terms if it solves a specific problem (e.g. compliance plus convenient access).

When cloud, when on-prem - concrete scenarios

Cloud usually wins when:

  • you are a small or medium business without an IT department
  • you want a deployment in 2-4 months, not in 8
  • you have multiple sites and want to see all of them in one system
  • you don't have the CAPEX for hardware but you accept operating costs
  • the business changes dynamically - the user count grows and shrinks

On-premise usually wins when:

  • you have an industry regulation forcing data control (pharma, parts of defence, certain government integrations)
  • the internet link at the location is bad and cannot be fixed
  • you have a strong IT team for whom server admin is a small extra
  • a perpetual licence fits your CAPEX/depreciation strategy
  • you are buying a WMS for 10+ years with no planned feature growth

Summary

The choice between cloud and on-premise in 2026 in a European SMB is rarely a technical decision, more often an organisational and capital one. The five-year price comes out roughly the same - so the call should depend on who prefers running infrastructure and what your industry actually looks like.

Weaver WMS is available in both variants - hybrid clients work with us alongside strictly cloud clients and clients with dedicated servers in rooms behind the office. For most SMBs we recommend the cloud, because it makes life simpler. But we don't pretend it is the only sensible choice - we see clients for whom on-premise simply fits better, and they should never have walked away from it. The decision starts with the question "what does your company want to run itself, and what does it not", and everything follows from there.