Skip to content

MES Architecture: On-Prem vs Cloud vs Hybrid

MES Architecture: On-Prem vs Cloud vs Hybrid
By Christian Fieg · Last updated: April 2026

TL;DR: Every MES runs on one of three architectures: On-Premise (local servers, full control, 12–24 month rollout), Hybrid / Lift-&-Shift (legacy MES hosted on cloud VMs, 6–12 months), or Cloud-Native (SaaS, microservices, 2–8 weeks). The architecture you choose determines your cost model (CAPEX vs OPEX), your scaling speed (per-plant vs per-click), your IT burden (dedicated team vs zero maintenance), and whether you can add new plants in days or months. For 80 % of midsize discrete manufacturers, cloud-native is the right answer in 2026. The 20 % exception: fully validated GMP/GxP production or air-gapped defense environments.

Transparency note: SYMESTIC is a 100 % cloud-native MES, hosted on Microsoft Azure. We do not offer an on-premise option. This article compares all three architectures honestly — including where on-premise is the right choice.

Table of contents

  1. What are the three MES architectures?
  2. How do they compare in practice?
  3. Which architecture fits your plant?
  4. What does a cloud-native MES architecture look like?
  5. How do you migrate from on-premise to cloud?
  6. FAQ

What are the three MES architectures?

An MES connects ERP-level planning (ISA-95 Level 4) with shop floor execution (Level 3). How that connection is built — where the software runs, how it scales, and who maintains it — defines the architecture.

1. On-Premise MES — The MES runs on servers in your own data center. Your IT team installs, maintains, patches, and backs up everything. All data stays on-site. Typical vendors: MPDV Hydra, Cronetwork (Industrie Informatik), GFOS. Rollout: 12–24 months. Cost model: CAPEX (license + hardware + consulting).

2. Hybrid MES (Lift & Shift) — An existing on-premise MES is moved to cloud-hosted virtual machines (e.g., Azure VMs). The software itself doesn't change — it's still monolithic. The hosting moves from your server room to a data center. This reduces local IT burden but doesn't unlock cloud-native benefits (auto-scaling, automatic updates, microservices). Rollout: 6–12 months. Cost model: mixed CAPEX/OPEX.

3. Cloud-Native MES — The MES is built for the cloud from day one. Microservices architecture. SaaS delivery. Automatic updates. No local infrastructure. Native APIs for ERP, IIoT, and analytics. Typical vendors: SYMESTIC, Tulip, 42Q. Rollout: 2–8 weeks. Cost model: OPEX (subscription per machine/plant).


How do they compare in practice?

Criterion On-Premise Hybrid (Lift & Shift) Cloud-Native
Hosting Your data center Cloud VM (Azure, AWS) SaaS platform
Architecture Monolithic Virtualized monolith Microservices
Cost model CAPEX (license + HW + consulting) Mixed CAPEX/OPEX OPEX (subscription)
Year-1 cost (10 machines) € 150 k–400 k+ € 100 k–250 k € 15 k–50 k
Implementation time 12–24 months 6–12 months 2–8 weeks
Adding a new plant New server, new license, new project New VM, configuration New database — done in days
Updates Manual (IT-planned downtime) Semi-automatic Automatic (zero downtime)
IT staff required Dedicated MES admin Shared IT resource Zero — vendor manages
IIoT / AI readiness Retrofittable (costly) Partial Native (APIs, MQTT, OPC-UA)
ERP integration Custom middleware Same middleware, remotely REST API + pre-built connectors
Data sovereignty Full on-site control Cloud provider region Cloud provider region (EU available)
Best for GMP/GxP, air-gapped, defense Transition phase Midsize discrete manufacturers

Which architecture fits your plant?

Your situation Recommended architecture Why
GMP/GxP-validated production, air-gapped network mandatory On-Premise Regulatory requirements mandate full local control. Cloud is not an option.
Existing on-prem MES, budget approved for infrastructure modernization only Hybrid (Lift & Shift) Buys time. Reduces local HW burden. But: plan the next step — this is a bridge, not a destination.
No MES yet. 1–6 plants. Discrete manufacturing. Want results in weeks. Cloud-Native No CAPEX, no IT project, no server room. Start on one line, scale to all plants.
Legacy MES is end-of-life. Full replacement needed. Cloud-Native Don't replace one monolith with another. Go cloud-native and eliminate the maintenance burden entirely.
Multi-plant enterprise, IT mandates SAP/Siemens ecosystem On-Premise or Hybrid (corporate mandate) — but evaluate cloud for non-core plants Corporate IT may require Opcenter or DELMIA. For satellite plants with less IT support, cloud-native is faster and cheaper.

What does a cloud-native MES architecture look like?

Layer What it does SYMESTIC implementation
Shop floor connectivity Reads machine signals via OPC-UA, MQTT, digital I/O IXON IoT devices, DI gateways, OPC-UA Cloud Connector
Data ingestion & transformation Timestamps, buffers, normalizes signals into a unified data model Azure IoT Hub → SYMESTIC Data Transformation layer
Application layer (MES) OEE, order tracking, alarms, SPC, SFM dashboards Modular: MDE, BDE, Kennzahlen, Fertigungssteuerung, Alarme, Prozessdaten
Integration layer Bidirectional ERP connectivity REST API + SAP ABAP IDoc + InforCOM + Navision file interface
User layer Dashboards, mobile app, shopfloor clients Browser-based — unlimited users, unlimited dashboards, smartphone MES app

Carcoustics — cloud-native at enterprise scale: 500+ machines across 7 countries (DE, PL, SK, CZ, MX, US, CN). IXON IoT devices → MQTT → Azure. Bidirectional SAP R3 integration via ABAP IDoc. Replaced a legacy on-premise solution in Poland and Haldensleben. All 7 plants connected within 6 months — not 24 months as the old system required. Results: 4 % fewer stoppages, 3 % higher output, 8 % better availability. Zero local server infrastructure at any plant.

Meleghy Automotive — 6-plant rollout: Started at Wilnsdorf. Scaled to Gera, Brandýs (CZ), Bernsbach, Reinsdorf, Miskolc (HU) within 6 months. Bidirectional SAP R3 + CASQ-it integration. SYMESTIC's modular architecture allowed Meleghy to scale independently — no SYMESTIC consulting needed for plants 3–6. Results: 10 % fewer stoppages, 7 % higher output, 5 % better availability.


How do you migrate from on-premise to cloud?

Phase What happens Duration
1 Pilot on one line. Connect 1–2 machines to the cloud MES. Compare data with the legacy system. Prove accuracy + speed. 2–4 weeks
2 Expand within the pilot plant. Connect remaining machines. Build dashboards. Integrate ERP. Train key users. 4–8 weeks
3 Parallel run. Run cloud MES alongside legacy system. Validate data consistency. Build confidence. 4–8 weeks
4 Decommission legacy. Scale to next plant. Switch off old system. Replicate configuration to plant 2, 3, 4… Ongoing

The critical insight: you do not need to migrate all plants at once. A cloud-native MES allows **plant-by-plant rollout** with independent timelines. Carcoustics started with a POC at Haldensleben and reached all 7 plants within 6 months — each plant at its own pace.


FAQ

What are the three MES architectures?
On-Premise (local servers, full control), Hybrid / Lift & Shift (legacy MES on cloud VMs), and Cloud-Native (SaaS, microservices, automatic updates). Each has different cost models, implementation timelines, and scaling characteristics.

Is a cloud MES secure enough for manufacturing?
Yes. Cloud-native MES platforms run on certified infrastructure (Azure, AWS) with enterprise-grade security (ISO 27001, SOC 2, encryption at rest and in transit). Data sovereignty is controlled by choosing the cloud region (EU data centers available). The IoT gateway uses outbound-only connections — no inbound ports opened.

How long does it take to implement a cloud MES?
First machines connected: 1–2 days. Full pilot plant with dashboards and ERP integration: 2–8 weeks. Multi-plant rollout: 3–6 months. Compare to on-premise: 12–24 months for a single plant.

Can I run cloud MES and on-premise MES in parallel?
Yes. Many manufacturers start the cloud MES on new lines or satellite plants while keeping the legacy on-premise system on validated lines. Over time, the cloud system expands and the legacy system is decommissioned.

What is the cost difference between on-premise and cloud MES?
On-premise: € 150 k–400 k+ in year 1 (license + hardware + consulting). Cloud-native: € 15 k–50 k in year 1 (subscription + onboarding). See our MES cost article for a detailed TCO breakdown.


The bottom line: MES architecture is not an IT decision — it is a business decision. On-premise gives control but costs time and money. Hybrid buys time but doesn't solve the scaling problem. Cloud-native delivers results in weeks, scales per click, and eliminates the IT burden entirely. For midsize discrete manufacturers with 1–6 plants, the architecture question in 2026 is answered: cloud-native — unless regulation explicitly prohibits it.

→ What is an MES? · → Cloud MES vs. On-Premise (Deep Dive) · → MES Costs · → MES Software Comparison · → IIoT · → OEE

About the author
Christian Fieg
Christian Fieg
Head of Sales, SYMESTIC · Previously iTAC, Dürr, Visteon (900+ connected machines) · Six Sigma Black Belt · LinkedIn
Start working with SYMESTIC today to boost your productivity, efficiency, and quality!
Contact us
Symestic Ninja
Deutsch
English